The Definitive Guide to Backup and Recovery for Cassandra EBOOK

The Definitive Guide to Backup and Recovery for Cassandra

Table of Contents

Executive Summary Executive Summary 1. Cassandra Technology Overview In today’s era of big data, enterprise applications 2. The Need for Backup & create a large volume of data that may be Recovery for Cassandra structured, semi-structured or unstructured in nature. Additionally, application development cycles are 3. Backup and Recovery much shorter and application availability is a critical Requirements requirement. Given these requirements, enterprises 4. Existing Practices for Backup are forced to look beyond traditional relational and Recovery databases to onboard next-generation applications (on IaaS or cloud-based PaaS). NoSQL databases 5. RecoverX Overview such as Apache Cassandra are now being adopted 6. Compare and Contrast and evaluated by enterprises for these applications, Cassandra Backup & including eCommerce, content management, etc. Recovery Solutions

7. Fortune 100 Retailer Customer Success Story

EBOOK 2 www.datos.io [email protected] Part I: Cassandra Technology Overview

Apache Cassandra™ is an open source, distributed and decentralized storage system (database) for managing large amounts of structured data across many commodity servers, while providing highly available service and no single point of failure. Apache Cassandra offers capabilities including continuous availability, linear scale performance, operational simplicity and easy data distribution across multiple data centers and cloud availability zones.

Key features of Cassandra include the following:

• Elastic scalability: Cassandra is highly scalable; more hardware can be added to accommodate more customers and additional data.

• Always on architecture: Cassandra has no single point of failure and is continuously available for business-critical applications.

• Fast linear-scale performance: Cassandra is linearly scalable, i.e., it increases throughput as a user increases the number of nodes in the cluster. Therefore, it maintains a quick response time.

• Flexible data storage: Cassandra accommodates all possible data formats including structured, semi-structured, and unstructured. It can dynamically accommodate changes to data structures.

• Easy data distribution: Cassandra provides the flexibility to distribute data by replicating data across multiple data centers.

• Transaction support: Cassandra supports properties like Atomicity, Consistency, Isolation, and Durability (ACID).

• Fast writes: Cassandra was designed to run on cheap commodity hardware. It performs fast writes and can store hundreds of terabytes of data, without sacrificing read efficiency.

Apache Cassandra’s architecture is responsible for its ability to scale, perform, and offer continuous uptime. All nodes play an identical role as each node communicates with each other equally. Apache Cassandra’s built-for- scale architecture is capable of handling large amounts of data and thousands of concurrent users or operations per second — even across multiple data centers — as easily as it can manage much smaller amounts of data and user traffic. Apache Cassandra’s architecture has no single point of failure and therefore is capable of offering true continuous availability and uptime — simply add new nodes to an existing cluster without having to take it down.

In Cassandra, one or more of the nodes in a cluster act as replicas for a given piece of data. If it is detected that some of the nodes responded with an out-of-date value, Cassandra will return the most recent value to the client. After returning the most recent value, Cassandra performs a read repair in the background to update the stale values.

EBOOK 3 www.datos.io [email protected] Part II: The Need for Backup & Recovery for Cassandra

As organizations increasingly rely upon NoSQL databases for their business-critical applications, and as these same organizations adopt a cloud-first strategy, the need for a comprehensive backup and recovery strategy has never been more critical. In this section, we will explore the need for a complete data protection strategy.

One of the advantages of Cassandra is that it provides scalability and high availability without compromising performance. Put simply, replication creates an alternative copy of an original and changes with the original in real time. So while native replication is good for protecting against media and network failures by replicating data across nodes, it can be a serious detriment in the case of data corruption or loss whereby the corrupted data set gets replicated, exacerbating an already bad situation.

Furthermore, the most common cause for data corruption and loss today is human error including operator error, fat finger mistakes, and accidental deletions. Recently, more malicious cyber threats have propagated including malware and ransomware. To protect against these threats, a comprehensive backup and recovery strategy is required. Unlike replication, backup is a copy of data taken at a certain point in time that does not change with time. A backup copy enables an administrator to restore data at a certain time in the past, typically a time before the data loss or ransomware attack occurred.

In summary, backup plays a critical role in an organization’s overall data protection strategy. In the next section, we’ll discuss in more detail the technical requirements for backup and recovery of Cassandra databases.

Part III: Requirements of Protection for Cassandra

This section will introduce the key requirements for protecting data that resides on Apache Cassandra, deployed either on-premise, or on private cloud with as-a-service model, or in public cloud with Amazon AWS, Google Cloud Platform.

Requirement #1: Online Cluster-Consistent Backups

One of the key requirements of next-generation applications that are deployed on Apache Cassandra is the always- on nature. This means that quiescing the database for taking backups is not feasible and moreover, the backup operation should not impact the performance of the application. As the application scales, the underlying Apache Cassandra also needs to scale-out to multiple shards. In this case, a backup solution must provide a consistent backup copy across shards without disrupting database and application performance during backup operations.

Requirement #2: Flexible Backup Options

Depending on the application, data may have different change rate and patterns. For example, in a product catalog, certain items may be refreshed everyday (fast selling goods), while the others may have longer shelf life (premium items). Based on the application requirements, some collections may need to be backed up every hour versus the

EBOOK 4 www.datos.io [email protected] others that may be backed up daily. Providing this flexibility to schedule backups at any interval and at collection level granularity is another requirement that we have heard from customers who are using Apache Cassandra. More importantly, these backups should always be stored on the secondary storage in native formats to avoid vendor lock-in.

Requirement #3: Handling Failure

Failures are a norm in the distributed database world. However, the backup solution should be resilient to database process failures, node failures, network failure and even logical corruption of data during backup and recovery operations. Finally, the backup solution should be able to handle failures of Apache Cassandra configuration servers that store metadata.

Part IV: Existing Solutions in the Marketplace

1. Traditional Backup and Recovery Solutions.

Traditional solutions were architected for IT applications deployed on traditional scale-up and static resources of compute and storage. They were designed for scale-up application environments with uncompressed workloads built to be operated within walls of on-premises infrastructure.

Media server based legacy backup architecture has no place in the cloud. Likewise, legacy media server based architectures have no place protecting next generation scale-out data.

Legacy backup solutions from long established backup vendors (e.g., Veritas, EMC/Legato, CommVault), are all based on the same legacy architecture: single vendor, end to end architectures which centralize the control of the backup process (the “control plane”) and moving/storing the backed-up data (the “data plane”). In these legacy backup solutions “media servers” act as the consolidated “control plane” and “data plane”.

Each backup vendor’s unique client “agent” software, installed on each host, used database specific APIs to invoke and process backups of structured data (e.g., Oracle RMAN). Likewise, client agents installed on each file server processed backing up unstructured files on each file server or used APIs on each NAS (an appliance based dedicated file server).

With an average architecture age of 20 years, legacy backup and recovery products from vendors including Veritas, Dell EMC, and CommVault were all designed before the advent of the cloud and modern applications, and are not architected to support the new, modern IT stack.

2. Manual Scripted Solutions

Manual solutions leverage native Apache Cassandra snapshot utility and scripts to transfer data to secondary storage. The scripts are customized for each Apache Cassandra cluster and require significant operational effort to scale or adapt to any topology changes. Further, these scripts are not resilient to failure scenarios e.g., failure

EBOOK 5 www.datos.io [email protected] of a node (primary or secondary) or intermittent network issues. Finally, recovery is a manual process, hence, time consuming and results in very high application downtime and contains data loss risk due to any bugs in the scripts. Overall, these solutions work when the Apache Cassandra environment is small and some data loss may be permitted in the application.

3. Backup and Recovery Via Replication

One of the longest running discussion threads in the backup world is explaining why “replication is not backup.”

In the early days of backup applications were built predominantly on relational databases and hosted on-premises in customers’ data centers. Users thought they were protected by creating copies of production data, on disk, replicated by disk arrays from a primary data site to a D/R site. If the primary site failed or went offline for whatever a reason, production could resiliently continue a D/R site. But replication quickly propagated data loss. Whatever the cause, if data becomes corrupted it gets replicated across multiple sites.

Whether it was before or after users suffered an unrecoverable data loss, customers eventually understood the reality that array-based replication enabled operational resiliency but was not “backup.” Backup solutions were required to deliver point in time, application consistent versions of data to enable rapid, point-in-time recovery in the event of data loss or corruption.

Next-generation scale-out databases do employ multi-node replication to guarantee operational resiliency. Even with the failure of one or more nodes in Apache Cassandra cluster transaction processing and application data access continues on, uninterrupted. But, next generation databases suffer the same inherent limitations of replication — accidental deletion of one or more data tables, fat finger errors, ransomware attacks all can result in unrecoverable data loss.

In fact, the clustered replication that makes next generation databases so resilient also makes them even more reliant upon backup to protect against data loss, corruption, and ransomware attack. But, traditional backup solutions that depend on silos of media servers and backup appliances, and store backups in a proprietary format on dedicated storage don’t address the application consistency and scalability requirements of next generation databases. Replication is NOT backup.

Part VI: RecoverX Overview

Rubrik Datos IO RecoverX: The Leading Solution for Scalable and Reliable Apache Cassandra Backup and Recovery

Rubrik Datos IO RecoverX is application-centric, data management solution providing backup and recovery for NoSQL, Non-Relational enterprise database systems. RecoverX delivers scalable versioning, single-click recovery, and industry-first semantic de-duplication, reducing redundant data and lowering storage costs. RecoverX includes any point-in-time backup and orchestrated recovery to safeguard against logical and human errors, malicious data corruption, application schema corruption, and other soft errors. Organizations can protect their data at any granularity and any point in time (flexible RPOs) to reduce application downtime with recovery in minutes

EBOOK 6 www.datos.io [email protected] (low RTOs), save up to 80% on secondary storage costs, and increase productivity of applications and DevOps teams. RecoverX enables continuous integration and development by fully automating the refresh of test and development environments using production data.

Advanced Features

• Advanced Recovery: The amount of data organizations need to protect is rapidly increasing but customers are constantly demanding faster RTOs to meet their business agility requirements and to address mandatory IT requirements such as those required by GDPR. These same organizations are typically building customer-centric applications that span hundreds of tables, with hundreds of TB’s of data (aka “Big Data”), with change rates approaching 300%, yet they expect recovery times of just a few hours. To address this new class of recovery requirements, RecoverX 2.5 now delivers advanced features including queryable recovery enabling fast, sub-table level recovery, as well as incremental and streaming recovery, both of which reduce recovery times down to minutes and minimize storage requirements for large scale restore operations.

• Backup Anywhere, Recover Anywhere: Cyber-attacks like ransomware present a real threat and organizations need to protect their data across data sources, applications, and any geographical boundary. Increasingly, customers have database tables spanning multiple data centers and multiple continents, but they want their data to be backed up and restored at the granularity of a single data center location. RecoverX 2.5 now delivers on this requirement with support for local backup of geo- distributed applications enabling local data center recovery for faster and more flexible RTO.

• Enterprise-Grade Security: Rubrik Datos IO is experiencing massive adoption in large enterprise customers across retail, ecommerce, financial services and healthcare organizations all of which have significant security requirements. RecoverX 2.5 introduces support for TLS/SSL encryption, support for X.509 certificates, LDAP authorization and Kerberos authentication to deliver optimized security between RecoverX and data sources and integration with enterprise security management tools.

Benefits

RecoverX provides organizations with the following benefits:

• Minimize application downtime through application consistent point-in-time backups by removing database repairs post recovery operations • Reduce secondary storage costs by ~80% using industry-first global semantic deduplication • Failure resiliency and elastic performance via highly-available software-only product • Operational efficiency through fully orchestrated recovery • Lightweight deployment in public cloud or private datacenter

EBOOK 7 www.datos.io [email protected] Part VII: Compare and Contrast: How RecoverX Stacks up for Cassandra Backup & Recovery

How Rubrik Datos IO Stacks up for Cassandra Backup and Recovery

Script Based Native Managed Rubrik Datos IO Customer Value & Solution Backup Solution RecoverX Benefits

Enterprise-grade backup and recovery Cluster Consistent Not Consistent - Not Consistent - versioning at Reduces Recovery Versioning/Restore database repair after database repair after any interval and time 2-3x - need for restore restore granularity database repair after restore lowest RTO, reduced restore time

Granular Backup Backup Granularity Key Space Key Space Column Family Policy versioning

Data From Cassandra Nodes to Generate Reduced time to Full Full Incremental complete backup to Subsequent Backup protection storage Versions

Parallel Backup of All Reduced time to Cassandra Nodes to No No Yes complete backup to Protection Storage protection storage

Yes, source node Failure Resiliency and Failure Handling None None failures handled reduced enterprise without data loss risk

Protected data is always de-duplicated

Yes, semantic Protected data is Backup Storage de-duplication for always stored on cost None None Capacity Reduction all backup version effective secondary storage choices storage

Up to 90% Storage Savings

Yes, Apache Multi-platform Support for Multiple Cassandra, DataStax None None enterprise-grade data Enterprise and Databases Platforms protection software MongoDB

EBOOK 8 www.datos.io [email protected] Part VIII: Customer Case Study (Fortune 100 Home Improvement Retailer)

The Rubrik Datos IO team listens. Over the last 9

Our Customer Achieved: months, we have given them a number of product “features, and the team delivers at a record pace. The Customer

Our customer is a Fortune 100 home improvement retailer” with thousands of brick and mortar stores in the U.S., Canada and 1 Hour Mexico. In addition, they have an eCommerce application that is central to their digital transformation journey. Their hyper-scale application is built with a cloud-first methodology that is deployed Using RecoverX, our natively on Google Cloud Platform (GCP). customer was able to backup multiple Cassandra The Need databases and recover from data loss with an SLA Our customer’s challenges can be summarized in three broad of 1 hour. In addition, our categories. customer increased their compute and memory The first isBackup and Recovery for Cassandra in Google Cloud. resources during peak The company is in the midst of their digital transformation journey seasons for seasonality, and and as their online business grew, they experienced challenges was able to get the required in scaling their existing relational databases. They re-architected backup performance. their online business application using a microservices based cloud native application architecture. This new application is deployed in

Google Cloud Platform, and uses underlying Cassandra (DataStax) databases to populate information for different customer-facing platforms. Given that their core online business is based on this webscale eCommerce application, any data loss is detrimental for their business. Our customer requires the ability to backup multiple Cassandra databases and recover from any data loss with an SLA of 1 hour.

The second category is Backup Software Elasticity. Being in the e-commerce industry, our customer experiences huge spikes in data volume during the holidays, especially Thanksgiving and Christmas. They could not rely on a static infrastructure footprint, and instead, wanted to scale resources only during peak seasons, and reduce the resources back after the season ended. This would ultimately allow them to optimize the cost of operating a data protection solution, without sacrificing high-availability.

The third category is Test Cluster Refresh Using Production Data. Our customer has multiple production and test clusters, and they implement continuous integration and continuous development (CI/CD) methodologies. Initially, they had to spend a considerable amount of time refreshing their test clusters because they are of different topologies.

EBOOK 9 www.datos.io [email protected] They needed a data management solution that would automate the refresh of test clusters using production data at regular intervals.

Key Challenges Why Rubrik Datos IO:

• Cloud-Native Backups of • The only cloud-native data protection solution for Cassandra (DataStax Enterprise) Apache Cassandra and DataStax Enterprise in Google Cloud • Data protection SLA of 1 hour and minimize application downtime • Cluster consistent backups that don’t require repairs on recovery, leading to low RTO 80% reduction in • Automate refresh of test secondary storage costs clusters using production data at regular intervals • High-performance to guarantee 1 hour SLA using elastic compute/storage resources

The Solution

Initially, our customer’s DevOps team used database native tools, but could not achieve the enterprise level data protection features that they needed. To quantify the impact of this any outage, it is estimated to be hundreds of thousands of dollars in lost business.

Rubrik Datos IO RecoverX was deployed natively in Google Cloud. RecoverX software was deployed in a clustered configuration to achieve high availability. A single RecoverX cluster was deployed to protect as many as 6 Cassandra clusters with a 1-hour backup interval. The data was stored on Google Cloud Storage for cost effectiveness.

The Results

Using RecoverX industry-first features, the customer was able to achieve operational and capital cost savings. At the same time, they were able to meet the organizational goal of 1-hour SLA.

Semantic deduplication resulted in 82% storage cost savings. Further, customer optimized its infrastructure costs by allocating appropriate compute resources depending on their application requirements.

About Rubrik Datos IO

Rubrik Datos IO is the application-centric data management company for the multi-cloud world. Our flagship Rubrik Datos IO RecoverX delivers a radically novel approach to data management helping organizations embrace the cloud with confidence by delivering solutions that protect, mobilize, and monetize their data — at scale. Rubrik Datos IO was recently awarded Product of the Year by Storage Magazine, and was recognized by Gartner in the 2016 and 2017 Hype Cycle for Storage Technologies. Rubrik Datos IO is headquartered in Palo Alto, California.

EBOOK 10 www.datos.io [email protected] The Definitive Guide to Backup and Recovery for Cassandra www.datos.io | 408.708.4136 | [email protected] 1001 Page Mill Rd, Bldg 2, Palo Alto, CA 94304