Storage MANAGING and PROTECTING ALL ENTERPRISE DATA

Storage MANAGING AND PROTECTING ALL ENTERPRISE DATA

EDITOR’S NOTE/RAFFO How to catch the NVMe wave

SNAPSHOT Cloud service disruptions and expectations

IT PRIORITIES Storage spending targets cloud and flash in 2018

STORAGE REVOLUTION/TOIGO Survive the coming unstructured data deluge Unprecedented, rapid commoditization of AFAs HOT SPOTS/SINCLAIR IoT and the shift toward Declining all-flash array prices are driving new technologies, SDS and the cloud market forces and opportunities. Here’s how you can benefit.

MAY 2018, VOL. 17, NO. 3

EDITOR’S LETTER DAVE RAFFO

Home

Raffo: Catch the NVMe wave How to catch should look at NVMe as the beginning of a transition to Unprecedented, storage-class memory (SCM). Switching to NVMe drives rapid AFA commoditization the NVMe wave makes for a relatively easy transition, and you will get a To prepare for NVMe, you must performance boost, although SAS SSDs meet the perfor- Snapshot: Cloud consider what’s coming next. mance requirements of the vast majority of applications. service disruptions and expectations

LOOKING PAST NVME TECHNOLOGY Storage spending targets cloud and It’s hard to look at NVMe without considering other new flash in 2018 technologies that will follow. The most obvious link is to NONVOLATILE MEMORY EXPRESS is probably on your radar, NVMe over Fabrics, but SCM technologies, such as Intel’s Toigo: Survive whether you’re already using or considering buying flash 3D XPoint Optane and Samsung’s Z-NAND media, should the coming unstructured storage. If you’re talking to a storage array vendor, you’re also be considered. data deluge almost certainly hearing all about the new host controller We already see NVMe technology in servers, and it will interface and storage protocol. eventually be common in storage arrays, hyper-converged Sinclair: IoT and the shift toward As vendors will tell you, nonvolatile memory express appliances and other forms of software-defined storage. SDS and the cloud is an inevitable step in enterprise flash. They will also tell For all storage array vendors’ talk about NVMe, few array you their NVMe products provide the best performance options are available with it yet. Those will come soon, About us and lowest latency, without disrupting the way you work. but, meanwhile, you should plan your next steps care- The part about performance and latency is true. NVMe’s fully. As with many new technologies, it will take time for bandwidth and improved queuing make NVMe SSDs bet- applications to support NVMe’s increased performance. ter performers than the SAS SSDs commonly used today. Consider what your storage tiers will look like when SCM, As far as which vendor’s NVMe product is the best, that dynamic RAM and various types of SSDs become storage has little to do with NVMe itself, just as you don’t generally options along with old-fashioned hard drives. judge flash arrays by comparing vendors’ drives. NVMe will also change the vendor landscape. Just as NVMe technology is part of an evolving flash world, the arrival of enterprise flash brought new vendors, such a step toward much more significant advances. You as Fusion-io, Kaminario, NetApp SolidFire, Nimbus Data,

STORAGE • MAY 2018 2 Pure Storage and Violin Systems, into play, there’s a new KEY QUESTIONS TO ASK Home batch of startups looking to ride NVMe into your data When you talk about NVMe with startups or established center. These include Apeiron Data Systems, E8 Storage, vendors, you must ask about their long-term strategies. Raffo: Catch the NVMe wave Excelero, Exten Technologies (formerly Mangstor), Pa- Look past your next array, as it will probably include vilion Data Systems, StorOne and Vexata. One or more of few changes beyond coming equipped with NVMe Unprecedented, these startups may come up with a better storage system SSDs instead of SAS SSDs. Ask vendors the following rapid AFA commoditization than your current array vendor. questions: Of course, established vendors are also staking their n Snapshot: Cloud short-term futures on NVMe technology. As with all-flash What do you mean by NVMe-ready? Is this merely re- service disruptions arrays, where only Pure cracked the big time as an inde- placing SAS SSDs with NVMe SSDs, or did you make and expectations pendent vendor, most of the startups won’t make it. But other architecture or management changes? you can bet Dell EMC, Hewlett Packard Enterprise, IBM, Storage spending n targets cloud and NetApp and Pure—now part of the establishment—will What is your plan for NVMe over Fabrics? Which fabrics flash in 2018 still be standing when the NVMe dust clears. will you support, and how will they affect the applications I’m running now or may run in the future? Toigo: Survive the coming unstructured n What is your SCM roadmap? How do you view future ad- data deluge Raffo recommends ... vances, such as 3D XPoint, Z-NAND and other emerging To be ready for the coming NVMe technology wave, technologies, and how can I prepare for them? Sinclair: IoT and I also recommend reading the following: the shift toward SDS and the cloud ■■ NVMe today and tomorrow The answers to these questions will help position you

■■ Storage class memory is persistent and your business to make good long-term choices when About us it comes to NVMe. n ■■ NVMe vs. SAS SSDs

DAVE RAFFO is editorial director of TechTarget’s Storage Media Group.

STORAGE • MAY 2018 3 ALL-FLASH ARRAYS

THE ALL-FLASH ARRAY market has seen a precipitous price decline over the past few years. AFA prices of $15 or $16 per gigabyte of raw capacity have fallen to $1 or less. Dis- counts of 80% or more are frequently offered. Storage buyers often believe they’re getting a good price from a big-name vendor, but then another vendor comes along at the last minute hungry for a deal and offers a better price. The AFA storage system price war is more surprising when considering how the NAND chip shortage of 2017 and early 2018 caused SSD pricing to increase during this same timeframe. What’s the root cause of the rapid commoditization of AFAs? A better understanding requires Unprecedented, some context.

rapid commoditization SKIMMING THE CREAM High-end products and services have always been subject of AFAs to commoditization. The first to market in a category Decline in all-flash array pricing provides new typically carries a premium price; that’s known as cream technologies, market forces and opportunities. skimming. As competition increases, the price gradually declines. These products and services use feature and BY MARC STAIMER brand differentiation to maintain higher pricing. But, eventually, prices decline as those features and functions show up in lower-priced competitors. PAVELVINNIK/GETTY IMAGESCREDIT HOME

STORAGE • MAY 2018 4 Differentiation is the perceived market value that products and services become commodities, price is the Home comes from a product’s ability to solve a specific problem, key differentiator. More sophisticated IT buyers may dif- as well as its attributes, features, quality, attractiveness ferentiate on TCO rather than price, although that, too, is Raffo: Catch the NVMe wave and brand. Commoditization happens when products the exception, not the rule. cease to be differentiated from their competitors. Once Data storage has been hit by this process. The storage Unprecedented, rapid AFA commoditization

Snapshot: Cloud Storage comparison metrics service disruptions and expectations PRICE PER GIGABYTE of raw capacity is the main metric protection capabilities, such as snapshots, replication, used by storage buyers to evaluate storage systems. mirroring, continuous data protection and high avail- Storage spending It commoditizes the storage buy in the following ways: ability; management software; and analytics software. targets cloud and flash in 2018 Price per raw gigabyte also fails to consider TCO, in- n Raw capacity is all of the drives’ rated capacity. cluding the costs of supporting infrastructure, mainte- Toigo: Survive nance, operations, upgrades, personnel to operate and the coming n Usable capacity is the amount of capacity left after the tech refresh data migration. It erroneously assumes all unstructured data deluge raw capacity has been formatted, file system imposed these values and costs are equal. and RAID established. More reliable metrics are TCO per effective usable Sinclair: IoT and gigabyte of capacity combined with TCO per random the shift toward n Effective usable capacity is the amount of capacity IOPS and TCO per sustained throughput. These met- SDS and the cloud available to be written to after thin provisioning, de- rics include actual costs that are substantially differ- duplication, compression, snapshots and clones are ent among competitive AFA products. They also include About us considered. hardware minimization effectiveness, which again var- ies by vendor product. It’s a flawed metric for several reasons. It doesn’t One area of note when assessing vendors: They will consider the value of performance; capacity minimi- vary on how much storage hardware minimization they zation, such as storage software efficiency, thin provi- guarantee because data deduplication and compression sioning, deduplication, compression and zero capacity will vary by data type. It’s prudent to get each vendor’s snapshots; performance and capacity scalability; data guarantee in writing. n

STORAGE • MAY 2018 5 pricing comparison metric has generally been price per gi- Magento, as well as infrastructure as a service, platform as Home gabyte of raw capacity, and, though imperfect, it’s still the a service and other as-a-service offerings. principal tool used to evaluate storage systems (see “Stor- When organizations run IT in the cloud, they don’t need Raffo: Catch the NVMe wave age comparison metrics”). One positive about this flawed in-house storage or an AFA. And even though running IT metric is it has steadily declined industrywide through the in the cloud costs more in the long term, it converts capital Unprecedented, years, mostly because of the continuous capacity gains of expenditures into operating ones while eliminating the rapid AFA commoditization both hard disk and solid-state drives. HDD capacity gains need to own and operate one or more data centers. This have slowed during the past few years; however, SSD gains has huge appeal to the midtier market and has caused

Snapshot: Cloud have accelerated as NAND fabrication has gone to 3D. many businesses to move more applications to the public service disruptions The steady decline in the gigabyte-of-raw-capacity metric cloud. The result is a shrinking total available market. and expectations doesn’t explain how AFAs got commoditized so fast. As established, commoditization results from increased Storage spending targets cloud and competition and an inability to differentiate value be- APPRECIABLY INCREASED DIRECT COMPETITION flash in 2018 tween competitive products. This is part of what’s hap- Why is there significantly more direct competition with pened in the AFA market. But there’s more, a lot more. all-flash array storage than there was previously with Toigo: Survive Three market trends have effectively caused the unprece- other storage system products? A lot of it is the result of the coming unstructured dented commoditization of AFA storage: substantial innovation in underlying storage technologies. data deluge The traditional gap between high-end storage sys- n a shrinking total available target market; tems—typically classified as enterprise or midtier—and Sinclair: IoT and n the shift toward appreciably increased direct competition; and mass-market or lower-end storage system counterparts SDS and the cloud n growing indirect competition. was once large in all respects. Lower-end systems couldn’t match high-end ones in performance, reliability, scal- About us ability, functionality and data protection, partly because SHRINKING TOTAL AVAILABLE TARGET MARKET high-end systems frequently had custom ASICs and highly Public cloud has been around for a decade. It’s had the engineered, complicated caching hardware systems to biggest effect on the small-to-medium business and mid- provide unmatched performance. And high performance tier markets. New businesses rarely have their own data was typically required in the shared storage environments. center. IT in the public cloud is simpler to set up and less As a result, there was also a large cost and price gap, pre- expensive upfront. It lets businesses take advantage of venting the commoditization of high-end systems. High prepackaged software-as-a-service applications, such as capacity, too, was exclusive to high-end systems. Microsoft Office 365, Google Suite, Salesforce, Oracle and Substantial technological advances have changed

STORAGE • MAY 2018 6 the old paradigm, narrowing and even eliminating the software stack to run on commercial off-the-shelf (COTS) Home performance and capacity gaps. The first advance to servers, also known as white box hardware. transform storage was the increasingly functional Intel Commodity hardware is a huge market shift. COTS Raffo: Catch the NVMe wave x86 microprocessor. Moore’s law may have slowed, but it server hardware eliminates the hardware premium pric- hasn’t stopped. The x86 CPU continues to become more ing once so prevalent in storage systems. Hard disk and Unprecedented, solid-state drives often had list prices 10-times higher in rapid AFA commoditization an enterprise storage system than the equivalent drive in COMMODITY HARDWARE a server. That drive in midtier storage could be six-times

Snapshot: Cloud IS A HUGE MARKET SHIFT. higher than the server equivalent. Even after all discounts, service disruptions COTS SERVER HARDWARE the drive in a server is seldom more than one-third the and expectations ELIMINATES THE ONCE price of the same drives in AFA storage systems. Also, maintenance costs of those drives after warranty is based Storage spending PREVALENT HARDWARE targets cloud and PREMIUM PRICING. on the higher Manufacturer’s Suggested Retail Price or list flash in 2018 price, not the net discounted price. SDS running on COTS server hardware means much Toigo: Survive powerful, adding more transistors, cores, power reduc- lower storage costs. Server manufacturers have been the coming unstructured tion and storage capabilities every couple of years. The aware of these technology changes and have developed data deluge latest iterations include several storage functions, such as a series of servers optimized for many drives, including XOR. These rapid CPU improvements make ASICs less all-flash designs. Sinclair: IoT and the shift toward attractive, with too little or no advantage compared to SDS and the cloud their upfront and ongoing costs, resulting in most storage systems now standardizing on x86 storage controller THE OPEN SOURCE ADVANTAGE About us architectures. Another major advancement has been open source storage The x86 platform architecture comes with loads of software, such as CentOS, Ceph, Docker, FreeBSD, Linux, standard off-the-shelf tools, making software develop- Swift and ZFS. It lowers the barriers to entry for new SDS ment faster and cheaper. It also enables the decoupling of or storage systems, because storage software developers storage software stacks from storage system hardware—a don’t have to start from scratch. Open source provides paradigm shift. This decoupling has allowed for the de- storage software features that are already developed and velopment of storage software separate from hardware, available at minimal costs. Even advanced functionality enabling a new class of competitor under the software-de- is available; for instance, deduplication is now part of fined storage, or SDS, banner. SDS enables the storage the Red Hat Enterprise Linux kernel. The storage system

STORAGE • MAY 2018 7 functionality barrier to entry has disappeared. competitors to surpass the fastest of the established Home Perhaps the biggest technology advancement and an es- enterprise storage systems, taking away their crown as sential element of all-flash array storage has been the flash the pre-eminent performance storage. (See “Upstart, Raffo: Catch the NVMe wave drive itself. SSDs have changed the storage performance high-performance AFA competitors” for a list of new envelope and leveled the playing field. Just about any AFA companies that have turned AFA performance into a Unprecedented, can now match or surpass the massive amounts of IOPS competitive advantage.) rapid AFA commoditization or throughput of enterprise AFA storage. IOPS latency, as with performance, is important to

Snapshot: Cloud many transactional applications. It, too, has seen signifi- THESE NIMBLE COMPETITORS service disruptions cant technological improvements with the development HAVE INCREASED PRICING and expectations of nonvolatile memory express SSDs and NVMe over PRESSURE ON BIG NAME Fabrics (NVMe-oF). Nonvolatile memory express is an Storage spending BRANDS, FORCING THEIR targets cloud and open standard driver that has significantly reduced latency PRICES DOWN. flash in 2018 between a server or controller and the NVMe drives, eliminating proprietary vendor driver stack lock-in in the Toigo: Survive process. NVMe-oF takes advantage of the NVMe driver Another innovation driving more AFA competition is the coming unstructured and remote direct memory access to provide similar low tiering, where cool or cold AFA data is archived to low- data deluge latencies over fabrics—Fibre Channel, Ethernet and er-cost secondary systems or cloud storage. Moving cool InfiniBand—to NVMe SSDs within a server or storage and cold data off of high-performance AFAs reduces the Sinclair: IoT and the shift toward system. Both are published standards and embraced by capacity required for primary data by 80% or more. That SDS and the cloud the open source community, as well as network adapter in turn lowers the cost of primary storage by an equivalent and SDS vendors. amount. Many legacy AFA storage vendors have adopted About us Using these advancements and standards, several new this capability, but frequently charge a fee to manage cool AFA vendors have become leaders in delivering much and cold data no longer on their arrays to make up for the lower latency, extreme IOPS and higher throughput. As loss of chargeable capacity. New and hungry competitors newcomers with less storage baggage, they’re more nimble don’t do that, giving them a competitive edge. and can take advantage of even newer storage advances The technological innovations and trends cited here such as storage-class memory drives—Intel Optane and have lowered the barriers to AFA market entry, creating a Micron QauntX 3D XPoint technology—which are faster slew of new competitors. These nimble competitors have and have lower latency than flash SSDs. Rapid adoption increased pricing pressure on big name brands, forcing of the latest storage technologies has enabled these new their prices down.

STORAGE • MAY 2018 8 GROWING INDIRECT COMPETITION into a single SKU and management infrastructure. Hy- Home The age of the IT specialist is in decline. With convergence per-converged infrastructure (HCI) goes deeper with the and hyper-convergence, the IT market pendulum has integration. It combines the server, hypervisor and storage Raffo: Catch the NVMe wave swung back toward integration of what have been separate within the server at a software level. Of the two types of disciplines of storage, networking and compute. convergence, HCI is growing faster because of lower costs Unprecedented, Converged infrastructure is the integration of the and simpler management. Nutanix is the HCI industry’s rapid AFA commoditization server, hypervisor, networks and storage hardware systems prime mover and sales leader, although most server and

Snapshot: Cloud service disruptions and expectations Upstart, high-performance AFA competitors

AFA TYPE: SYSTEM OR Storage spending VENDOR PRODUCT* OTHER PERFORMANCE ADVANTAGES SOFTWARE-DEFINED STORAGE targets cloud and flash in 2018 Apeiron Data Systems ADS Series System Scale-out NVMe-oF nodes of NVMe drives

Toigo: Survive E8 Storage software Scale-out NVMe-oF nodes of Optane the coming E8 Storage Both unstructured and appliance and NVMe drives data deluge Scale-out Remote Direct Drive Access Excelero NVMesh Both between nodes Sinclair: IoT and the shift toward SDS and the cloud Exten Technologies NX6325 System NVMe and NVMe-oF drives (formerly Mangstor)

About us Pavilion memory Pavilion Data Systems System Greater performance density per rack unit array

Highly efficient storage software collapses TRU (Total Resource StorOne Both storage stack, lowers latency and gets more Utilization) Storage performance from less hardware

Optane SSDs, parallel switching, separate data Vexata VX-100 Both and control path reducing system latency

*ALL SUPPORT NVME AND NVME OVER FABRICS TECHNOLOGIES; SOURCE: DRAGON SLAYER CONSULTING

STORAGE • MAY 2018 9 white box vendors now sell HCI systems. differentiate itself. It also covers its bases by selling mul- Home HCI doesn’t need stand-alone AFAs. It has redundant tiple products in each category of AFA, SDS, converged storage, and all-flash SSDs in HCI cost a lot less. It’s similar infrastructure and HCI. NetApp also sells products in each Raffo: Catch the NVMe wave to SDS, with the drives costing two-thirds or less than the category and guarantees quality of service for its SolidFire drives in AFAs. HCI, in effect, reduces the need for stand- array and HCI, as well as the scalability and snapshot ser- Unprecedented, alone, shared AFAs. vices of its ONTAP OS. rapid AFA commoditization All that differentiation helps vendors stand out and solve different problems while slowing commoditization

Snapshot: Cloud NEED FOR DIFFERENTIATION of their products—but only to a point. Differentiation service disruptions Commoditization occurs when there’s no perceived dif- doesn’t overcome the shrinking market and increased and expectations ferentiation in problems solved, value, functionality or competition pressures. brand. This has led several AFA vendors to attempt to Storage spending targets cloud and differentiate their products. flash in 2018 Some, such as IBM, Nimbus Data, Pavilion Data Sys- WHAT THIS ALL MEANS tems, Pure Storage and Western Digital, make unique SSD A shrinking market due to the rapid growth of the cloud Toigo: Survive form factors. This lets them increase their storage density and HCI plus the explosion of competitors will continue the coming unstructured per rack unit. Another group has tried to differentiate to force all-flash array pricing down and commoditization data deluge their AFA storage through performance, as mentioned up. Differentiation acts as a speed bump barely slowing earlier. Others have developed unique hardware archi- this race to the pricing bottom. This is good news for Sinclair: IoT and the shift toward tectures. Infinidat, for example, isn’t an AFA, but a hybrid storage buyers, but not storage vendors. Margins will SDS and the cloud array of SSDs and HDDs that claims AFA performance continue to shrink, competitors will disappear and new with up to 3 TB of dynamic RAM caching. leaders will emerge. About us Hewlett Packard Enterprise now relies primarily on The AFA storage vendors that survive will solve urgent analytics (Nimble) and less so on a unique ASIC (3PAR). and costly user problems that others don’t, or they’ll solve StorOne’s approach is based on its complete rewrite of them better than their competitors. Differentiation purely the storage stack and collapsing the many layers into a on price or cost is a race to the bottom that benefits no one single process or layer that requires much less hardware in the long term. n to provide the same IOPS and throughput. Dell EMC bundles additional software and services to MARC STAIMER is president and CDS of Dragon Slayer Consulting.

STORAGE • MAY 2018 10 Snapshot Home CLOUD SERVICE DISRUPTIONS AND EXPECTATIONS BY THE NUMBERS —James Alan Miller

Raffo: Catch the NVMe wave

Making the move to the cloud Misconceptions about cloud service Unprecedented, D D rapid AFA commoditization 100%

Snapshot: Cloud 99% service disruptions 80% and expectations 59% 60% Storage spending mistakenly believe their cloud service provider targets cloud and is primarily responsible for dealing with flash in 2018 40% cloud outages.

Toigo: Survive 30+ 20+ 50+G the coming 20% unstructured 27% data deluge

0 Sinclair: IoT and intend to expect to turn to 83% the shift toward transfer systems the public cloud SDS and the cloud to the cloud in to outsource incorrectly believe cloud service providers the next year all on-premises are primarily responsible for safeguarding workloads and data from cloud service disruptions. About us or two. infrastructure. 42+ 8+ 50+G

of respondents have yet to completely assess how 60% much a cloud outage will cost their businesses.

SOURCE FOR ALL CHARTS: “THE TRUTH IN CLOUD REPORT,” VANSON BOURNE FOR VERITAS TECHNOLOGIES, 1,200 GLOBAL BUSINESS AND IT DECISION-MAKERS SURVEYED, JULY-AUGUST 2017.

STORAGE • MAY 2018 11 Downtime per month due to cloud Home D outages

Raffo: Catch 40% the NVMe wave

Unprecedented, rapid AFA 30% Less than 15 minutes 31% commoditization Amount of downtime 36% of IT professionals 27% expect to experience from cloud service Snapshot: Cloud 20% interruptions each month service disruptions 20% 21% and expectations 22 minutes 10% Storage spending Average amount of downtime surveyed targets cloud and businesses actually experience from flash in 2018 cloud outages per month 0

Toigo: Survive >30 16 to 30 0 to 15 None the coming minutes minutes minutes unstructured data deluge D Effects of cloud service interruptions*

Sinclair: IoT and System downtime the shift toward D Has your organization been affected 73% SDS and the cloud by a cloud outage? 46% Lower customer satisfaction

About us Yes 37% Loss of revenue

26% Inability to test workloads No 6% 58% Don’t 25% Noncompliance with regulations 36% know 25% Fines and penalties

*MULTIPLE SELECTIONS ALLOWED

SOURCE FOR ALL CHARTS: “THE TRUTH IN CLOUD REPORT,” VANSON BOURNE FOR VERITAS TECHNOLOGIES, 1,200 GLOBAL BUSINESS AND IT DECISION-MAKERS SURVEYED, JULY-AUGUST 2017; CLOCK: PANIMONI/GETTY

29+ 18+ 3+ 50+p STORAGE • MAY 2018 12 IT PRIORITIES

IT’S HARDLY NEWS that surging data growth has put the squeeze on storage systems. To cope, organizations plan to spend heavily on cloud-based backup and the use of cloud and local flash for primary storage. Those are among the top IT spending priorities for data storage highlighted in the TechTarget 2018 IT Priorities Survey. The survey culled the results of 181 respondents in North America who identified data backup and storage as areas that dominate their time. These IT and storage pros work for organizations that maintain an average of 1.8 petabytes (PB) of capacity on premises. Twenty-one percent of these organizations have 1 PB Storage spending or more, including 8.5% whose storage exceeds 10 PB of capacity. Another 24% have between 250 TB and 999 TB, targets cloud while 34% have between 10 TB and 249 TB. About 20% have fewer than 10 TB. and flash in 2018 Increased data growth drives enterprise CLOUD STORAGE PRIORITIES priorities for primary storage and backup. The respondents’ average IT budget will increase 6% in BY GARRY KRANZ 2018, roughly the same as 2017. Fifty-four percent of enterprise IT budgets will rise this year, with 12% expected to increase more than 10%. Another 19% anticipate IT (Continued on page 15) MAKSIM KABAKOU/ADOBE STOCK HOME

STORAGE • MAY 2018 13 Home Top primary storage initiatives for 2018 vs. 2017 Raffo: Catch the NVMe wave 25% Hybrid storage arrays* NA Unprecedented, rapid AFA 25% All-flash arrays commoditization 31% 21% Snapshot: Cloud Public cloud* service disruptions NA and expectations 21% Hyper-converged infrastructure 24% Storage spending targets cloud and Converged infrastructure 21% flash in 2018 17%

Storage for virtual environments 20% Toigo: Survive 27% the coming unstructured Data management systems* 20% data deluge NA

Private, on-premises cloud* 18% Sinclair: IoT and NA n 2018 the shift toward n 2017 SDS and the cloud Storage virtualization 18% 16% About us Data reduction 18% 25%

Software-defined storage 13% 21%

Scale-out NAS 13% 17%

SOURCE: TECHTARGET 2017 AND 2018 IT PRIORITIES SURVEYS; *NOT INCLUDED IN 2017 SURVEY; MULTIPLE SELECTIONS PERMITTED.

STORAGE • MAY 2018 14 (Continued from page 13) would be spending on public and private cloud initiatives Home spending will grow between 5% and 10%. Twenty-three in 2017. Still, nearly one in five data centers will turn to percent of budgets will increase by less than 5%, 20% will the cloud for production workloads, while 21% expect to Raffo: Catch the NVMe wave remain flat and 7% will decrease. use public cloud providers. Eighteen percent will launch Cloud, backup storage and on-premises flash are among private, on-premises clouds. Unprecedented, the top IT spending priorities for data storage. Only 5% signaled their intentions to spend money to rapid AFA commoditization The number of companies planning to use cloud storage implement primary object storage, a surprisingly low for primary applications dropped at least 9 percentage number considering cloud pricing wars among the major

Snapshot: Cloud points, and possibly more, to 39% from 48% who said they (Continued on page 17) service disruptions and expectations

Storage spending targets cloud and flash in 2018 IT budget change 2017 vs. 2018

Toigo: Survive the coming unstructured 24% data deluge n 2018 24% 23% n 2017 21% 20% 19% Sinclair: IoT and the shift toward SDS and the cloud 12%

About us 6% 3% 3% 3% 4% 3% 1%

Decrease by Decrease by Decrease by No change Increase by Increase by Increase by more than 5% to 10% less than less than 5% to 10% more than 10% 5% 5% 10%

SOURCE: TECHTARGET 2018 AND 2017 IT PRIORITIES SURVEYS

STORAGE • MAY 2018 15 Home Top backup storage initiatives for 2018 vs. 2017 Raffo: Catch the NVMe wave 32% Backup for virtual servers 43% Unprecedented, rapid AFA 25% Snapshots or replication commoditization 29% 24% Snapshot: Cloud Continuous data protection service disruptions 27% and expectations In-house or internal business 24% continuity/DR 26% n 2018 Storage spending n 2017 24% targets cloud and Backup software refresh flash in 2018 19% 23% Data deduplication for backup Toigo: Survive 30% the coming unstructured 22% Application-specific backup* data deluge NA 21% Sinclair: IoT and Cloud backup the shift toward 48% SDS and the cloud 21% Laptop or mobile device backup 18% About us Cloud-to-cloud backup* 18% NA

Backup disk array 17% 26%

Outsourced/cloud business continuity/DR 13% (cloud DR or disaster recovery as a service) 16%

SOURCE: TECHTARGET 2017 AND 2018 IT PRIORITIES SURVEYS; *NOT INCLUDED IN 2017 SURVEY; MULTIPLE SELECTIONS PERMITTED.

STORAGE • MAY 2018 16 (Continued from page 15) DATA CENTERS EYE FLASH, HYPER-CONVERGED Home providers. Most object storage runs in the cloud, predomi- On-premises storage infrastructure is a priority at 37% of nantly using AWS S3. In 2017, 16% of those surveyed listed companies. And all-flash and hybrid arrays are running Raffo: Catch the NVMe wave object storage as a spending priority. neck and neck, each considered by 25% of companies in Other notable findings in primary storage include con- 2018. Unprecedented, tinuing demand for data management (20%) and data A new wrinkle in technology spending priorities for rapid AFA commoditization reduction (18%) tools. About 12% of respondents will 2018 is the emergence of nonvolatile memory express spend money to implement Fibre Channel storage or ex- (NVMe) flash storage. Although still in its early days,

Snapshot: Cloud pand an existing network, and the same percentage plans nearly 7% of data center administrators at organizations service disruptions to implement Ethernet-based storage. In addition, 9% of responding to the survey said they will invest in NVMe- and expectations respondents are set to launch open source initiatives in based flash this year. the coming year. Thirteen percent of storage teams have scale-out Storage spending targets cloud and A significant number of respondents placed cloud- NAS on their radar for 2018, down from 17% in 2017. flash in 2018 based backup and disaster recovery (DR) high on their Meanwhile, storage with converged systems continues to-do lists. That includes 21% with plans to back up local to gain attention in primary storage. Among our survey Toigo: Survive data to the cloud and 18% implementing maturing cloud- respondents, 38% will deploy hyper-converged infrastruc- the coming unstructured to-cloud backup technologies. Another 13% will allocate ture, compared with 23% who lean toward converged data deluge part of their storage budget to cloud-based business infrastructure, as a data center networking initiative. continuity and DR—although that dropped from 16% of Both hyper-converged infrastructure and converged in- Sinclair: IoT and the shift toward respondents a year ago. frastructure come in at 21% when the question is limited SDS and the cloud Data centers are also taking steps to handle the chal- to organizations using those technologies for primary lenge of managing multiple cloud environments, with storage initiatives, however. About us more than 30% on track to consolidate cloud workloads A hyper-converged platform combines computing, net- in 2018. working, storage and virtualization resources integrated Among other top IT spending priorities for 2018 in data as a single managed appliance. Converged infrastructure storage highlighted in the survey include the following: allows the system components to be separated from the Companies are continuing to de-emphasize in-house chassis and used individually. data protection. The number of companies spending on Emerging composable infrastructure, which uses soft- disk-based backup appliances is 17%, down from 26% and ware to assemble compute, network and storage resources in-house DR budgets dropped from 26% in 2017 to 24% as needed on bare metal, is a budgetary priority for nearly in 2018. 15% of data centers for their infrastructures.

STORAGE • MAY 2018 17 Thirty-seven percent of enterprises surveyed are of respondents cited backup of virtual servers as one of Home planning to invest in software-defined storage, includ- their top initiatives this year. About 20% will add storage ing server-side flash installations, for their data center for virtual environments. Fifty-six percent will embark on Raffo: Catch the NVMe wave infrastructures. server virtualization. Snapshot replication is also a big secondary storage Unprecedented, concern, with 25% of respondents planning initiatives in rapid AFA SECONDARY STORAGE COMES FIRST commoditization 2018, and 24% will spend storage dollars on continuous Aside from greater use of the cloud, IT blueprints for data protection technologies. n

Snapshot: Cloud backup encompass a variety of strategies and tools. Deal- service disruptions ing with storage for virtualization remains among the top GARRY KRANZ is a senior news editor covering data storage at and expectations IT spending priorities in data storage. Nearly one-third TechTarget.

Storage spending targets cloud and flash in 2018

Toigo: Survive the coming unstructured data deluge

Sinclair: IoT and the shift toward SDS and the cloud

About us

STORAGE • MAY 2018 18 STORAGE REVOLUTION JON TOIGO

Home

Raffo: Catch the NVMe wave How to survive sitions and improves workload and data productivity. This Unprecedented, is the thinking behind Mohit Aron’s latest storage venture, rapid AFA commoditization the unstructured Cohesity and its hyper-converged secondary storage.

Snapshot: Cloud data deluge service disruptions THE PROBLEM WITH NAS and expectations Cohesity responds to the shortcomings of NAS with a novel approach to secondary Data has grown at an accelerated rate over the last few years and is expected to continue to do so for the fore- Storage spending storage. targets cloud and seeable future. Unstructured data, both files and objects, flash in 2018 is a significant contributor to this growth and is usually directed to NAS or other file server configuration storage Toigo: Survive silos. There are many problems that mostly center on the coming unstructured complexity and cost. data deluge MANY IT MANAGERS look at data growth rates with trepida- With NAS platforms, vertical scalability is limited. You tion. For years, larger shops have hosted files and objects can only fill the crate with so many hard drives or SSDs. Sinclair: IoT and the shift toward on dedicated NAS platforms, using the latest technologies NAS is essentially a thin server bolted to an array, usually SDS and the cloud from vendors such as NetApp and Dell EMC Isilon to with an expensive memory buffer on a feature card to handle the growth. However, traditional NAS appliances spoof applications so they don’t see any latency in write About us aren’t scalable enough to consolidate proliferating second- operations. Once all the slots are filled, your only choice ary storage workloads, including file and object, backup is to deploy more appliances and scale horizontally. and archive, testing and development, and analytics. In addition to hardware, horizontal scaling requires These appliances are fast becoming silos that complicate more thin server software licenses. Moreover, to build a data sharing and make infrastructure management a Her- coherent, expanded silo, you may also need software to culean undertaking. help the individual appliances work in a group. And spe- We need a replacement architecture that delivers cialty software may be needed to establish and maintain scalability, reduces the management burden, bends the a global namespace over pooled hardware so you can see capacity-demand curve to slow the need for capacity acqui- files in a one-throat-to-choke manner. There are many

STORAGE • MAY 2018 19 shortcomings to this strategy, with several ways to break node model that’s managed as a single set of resources Home the storage, including corrupted namespaces, pooled to host all data from secondary storage workloads that hardware failures and failover errors. It also has a cost are fragmented and siloed in traditional NAS storage Raffo: Catch the NVMe wave model that grows unpredictably, because of differences models. This simplifies both the management of the in storage architecture, topology and protocols over time. infrastructure and the access to data stored there. Unprecedented, rapid AFA n Web-scale. commoditization Cohesity hyper-converged secondary storage AN ALTERNATIVE ARCHITECTURE features unlimited scalability with universal deduplica-

Snapshot: Cloud Cohesity started as an effort to solve some of the woes of tion. It’s among the first with global, variable, block-level service disruptions backup, said Aron, the company’s CEO. The goal was to data reduction. Start as small as three nodes and scale and expectations establish a secondary storage platform that could support up at will without a lot of hassle. unlimited web-scale capacity—both on premises and in Storage spending targets cloud and clouds—without a lot of administrative or managerial in- n Productivity. You can share data among workloads di- flash in 2018 tervention. A fan of deduplication, Aron wanted to make rectly rather than requiring data-copy sharing. Global this and other functionality available across the secondary indexing and searching helps you find the data you Toigo: Survive storage platform rather than as value-add software on need, and in-place analytics supports queries and trend the coming unstructured individual appliances or arrays. analysis, while also improving the ability of different data deluge Aron is a veteran of Nutanix, which did something workloads to use data. similar with primary storage, using software-defined Sinclair: IoT and n Multi-cloud. the shift toward storage and hyper-converged infrastructure technology Cohesity’s hyper-converged secondary SDS and the cloud and architecture to provide a platform for mission-critical storage can span from the data center’s edge to multiple workloads. As a developer of the Google File System, Aron public and private clouds to capitalize on cloud elasticity About us had the right stuff to create Cohesity’s hyper-converged and economics, while using data wherever it’s placed in secondary storage platform. the infrastructure. The company touts the benefits of hyper-converged secondary storage for various applications, including file and object storage. And the platform does seem to be an CHECKING ALL THE RIGHT BOXES improvement over traditional NAS in the following ways: Cohesity is proud of its SpanFS distributed file system, the root of the secondary storage platform’s advanced func- n Hyper-convergence. Cohesity’s platform is hyper-con- tionality. The company’s consistency guarantee, where verged, meaning you’re dealing with a clustered storage its hyper-converged secondary storage writes to multiple

STORAGE • MAY 2018 20 nodes before acknowledging a write, is particularly note- boxes. Plus, the product is sold on a pay-as-you-grow ba- Home worthy. This approach is different from NAS products, sis, eliminating the forced forklift upgrade and warranty which deliver eventually consistent writes, where data is renewal models of NAS vendors. Raffo: Catch the NVMe wave written to a cache until it finally writes to the back-end It’s clear, out-of-the-box thinking is required to cope storage and can result in data loss if certain interruption with the looming unstructured data deluge. Rather than Unprecedented, events occur. bolting on additional features or functions to an appliance rapid AFA n commoditization Add fully linear performance scaling and a bunch of model, Cohesity’s approach is worth a look. data protection features—such as erasure coding, including ratios of 2-to-1, 4-to-2 and 5-to-2 and replication fac- Snapshot: Cloud JON WILLIAM TOIGO is a 30-year IT veteran, CEO and managing service disruptions tor—to the functionality story, and it appears Cohesity’s principal of Toigo Partners International, and chairman of the Data and expectations file and object storage platform checks all the right tech Management Institute.

Storage spending targets cloud and flash in 2018

Toigo: Survive the coming unstructured data deluge

Sinclair: IoT and the shift toward SDS and the cloud

About us

STORAGE • MAY 2018 21 HOT SPOTS SCOTT SINCLAIR

Home

Raffo: Catch the NVMe wave IoT and the according to Enterprise Strategy Group’s latest IT spend- Unprecedented, ing intentions data. This means 25% of companies already rapid AFA commoditization shift toward SDS have programs in place to collect data to become more op- erationally efficient; better monitor products and services; Snapshot: Cloud and the cloud deliver a superior customer experience; and even develop service disruptions new products, services or business models. and expectations Emergent workloads, like the internet of things, are forcing changes in data center Unlike other workload categories, it’s important to note that IoT isn’t a single thing. It’s a collection of hundreds Storage spending architectures. targets cloud and and even thousands of industry- or project-specific initia- flash in 2018 tives and workloads. This is an important distinction. For example, IoT projects may involve collecting and analyz- Toigo: Survive ing data from “smart” consumer products, or environmen- the coming unstructured tal monitoring as part of a smart city initiative, or factory data deluge TRANSFORMATION HAS BEEN the main narrative in IT lately. floor data to improve collaboration among machines and With the rise of the digital economy, the connection people. However, in spite of this diversity, common trends Sinclair: IoT and the shift toward between data and business results is stronger than ever. have emerged in how IT organizations respond to the SDS and the cloud The competitive drive to use data more effectively to increased data storage and access demands IoT generates. push business results, revenue and profitability has led to Evidence suggests a link between IoT and software-de- About us the popularity of newer workloads that are changing the fined storage (SDS) and IaaS public cloud options. In the nature of IT. These workloads include analytics, machine Enterprise Strategy Group’s research, IT shops that indi- learning, blockchain and the internet of things. It’s a trans- cated IoT as a top driver of data growth were more likely to formation different from any we’ve seen before and with be committed to SDS as a long-term strategy. Additionally, serious repercussions for those who don’t adapt. more than 75% of companies with IoT initiatives under- The internet of things offers an excellent illustration way use IaaS. Where IoT was a top driver of data growth, of how emergent workloads are forcing changes in data there was a higher likelihood of using IaaS. SDS and IaaS center architectures. A quarter of IT decision-makers in- offer several common benefits that make sense for IoT dicated their organizations have IoT initiatives underway, environments. These include the following:

STORAGE • MAY 2018 22 n Scale. IoT projects tend to generate tremendous amounts efforts, such as internet of things initiatives. Home of data. Using analytics at the edge, data that’s sent to and retained in the data center can be significantly reduced, n Pay per consumption. The flipside of infrastructure Raffo: Catch the NVMe wave but, in aggregate, IoT initiatives are designed to capture scalability is controlling the cost of that scale. Not sur- a huge amount of data that will scale over time. Here’s prisingly, the rate of data growth and the cost of storage Unprecedented, where SDS offers benefits. Designed to be hardware-ag- infrastructure continue to be among the top three most rapid AFA commoditization nostic, SDS technologies aren’t typically confined by ar- commonly identified storage-related challenges. Cloud bitrary capacity limits and often support large capacities. services, such as IaaS, are known for their pay-per con-

Snapshot: Cloud IoT workloads also tend to generate unstructured data sumption models. Many SDS products offer similar service disruptions sets, ideal for the scalable file and object storage com- payment structures as well. By only paying for what and expectations monly offered as SDS technologies. Similar comments can you consume, you can reduce infrastructure costs. The be made for IaaS, with its ability to scale data capacity as benefits, however, extend further. Pay-per-consumption Storage spending targets cloud and needed and as quickly as required. models also make it easier to adjust the pace of storage flash in 2018 consumption if demand becomes less predictable. That n Agility and flexibility. Hardware flexibility, often an can make it easier to consume infrastructure faster than Toigo: Survive inherent component of SDS and IaaS architectures, con- anticipated while still optimizing costs. There are, of the coming unstructured tributes significantly to scalability. This flexibility is about course, differences in the cost structures among various data deluge more than just scalability, however, as SDS architectures IaaS and SDS services and products. These added costs, typically deliver consistent data access while the underly- such as for data egress common to IaaS, must be consid- Sinclair: IoT and the shift toward ing hardware changes, evolves and adapts. Hardware flex- ered in any infrastructure decision. SDS and the cloud ibility lets you integrate new hardware capabilities, such as faster memory and processing, without having to wait These three capabilities should be part of any IT storage About us for a full system refresh. This approach helps reduce the infrastructure modernization effort and not just for IoT overall cost of infrastructure. Furthermore, neither SDS initiatives. And while IaaS is often associated with these nor IaaS generally requires generational forklift upgrades. three capabilities—and rightly so—increased interest in This reduces both the infrastructure and management SDS reminds us there are multiple on-premises storage costs of storage, as well as the potential risk of a migration products that deliver cloud-like benefits as well, such as procedure. Software-defined storage architectures also greater agility, flexibility and scale.n tend to improve infrastructure agility, the speed at which

new capacity can be deployed and provisioned. This in SCOTT SINCLAIR is a senior analyst with Enterprise Strategy Group in turn speeds up the delivery of new data-driven business Austin, Texas.

STORAGE • MAY 2018 23 Home

Raffo: Catch the NVMe wave Storage is a SearchStorage.com e-publication. Unprecedented, rapid AFA Dave Raffo, Editorial Director commoditization James Alan Miller, Executive Editor, [email protected] Snapshot: Cloud service disruptions Stacey Peterson, Managing Editor and expectations George Crump, Scott Sinclair, Marc Staimer, Jon Toigo, Contributing Editors Storage spending targets cloud and Linda Koury, Director of Online Design flash in 2018 Nick Arena, Associate Managing Editor, E-Products Toigo: Survive the coming Jillian Coffin,Publisher, [email protected] unstructured data deluge SEARCHSTORAGE.COM SEARCHCONVERGEDINFRASTRUCTURE.COM Sinclair: IoT and SEARCHDATABACKUP.COM the shift toward SEARCHDISASTERRECOVERY.COM SDS and the cloud

About us Stay connected! SearchStorage | Facebook | @SearchStorageTT | LinkedIn

© 2018 TechTarget Inc. No part of this publication may be transmitted or reproduced in any form or by any means without written permission from the publisher. TechTarget reprints are available through The YGS Group. About TechTarget: TechTarget publishes media for information technology professionals. More than 100 focused websites enable quick access to a deep store of news, advice and analysis about the technologies, products and processes crucial to your job. Our live and virtual events give you direct access to independent expert commentary and advice. At IT Knowledge Exchange, our social community, you can get advice and share solutions with peers and experts.

COVER IMAGE AND PAGE 4: PAVELVINNIK/GETTY IMAGES

STORAGE • MAY 2018 24