Backup and Disaster Recovery in Hadoop
•Download as PPTX, PDF•
10 likes•6,605 views
From: DataWorks Summit Munich 2017 - 20170406 While you could be tempted assuming data is already safe in a single Hadoop cluster, in practice you have to plan for more. Questions like: "What happens if the entire datacenter fails?, or "How do I recover into a consistent state of data, so that applications can continue to run?" are not a all trivial to answer for Hadoop. Did you know that HDFS snapshots are handling open files not as immutable? Or that HBase snapshots are executed asynchronously across servers and therefore cannot guarantee atomicity for cross region updates (which includes tables)? There is no unified and coherent data backup strategy, nor is there tooling available for many of the included components to build such a strategy. The Hadoop distributions largely avoid this topic as most customers are still in the "single use-case" or PoC phase, where data governance as far as backup and disaster recovery (BDR) is concerned are not (yet) important. This talk first is introducing you to the overarching issue and difficulties of backup and data safety, looking at each of the many components in Hadoop, including HDFS, HBase, YARN, Oozie, the management components and so on, to finally show you a viable approach using built-in tools. You will also learn not to take this topic lightheartedly and what is needed to implement and guarantee a continuous operation of Hadoop cluster based solutions.
Report
Share
More Related Content
What's hot
What's hot (20)
Similar to Backup and Disaster Recovery in Hadoop
Similar to Backup and Disaster Recovery in Hadoop (20)
More from larsgeorge
More from larsgeorge (13)
Recently uploaded
Recently uploaded (20)
Backup and Disaster Recovery in Hadoop
- 1. Backup and DR in Hadoop Lars George – Partner and Co-Founder @ OpenCore DataWorks Summit Munich 2017 Distributed Problems
- 2. About Me • Partner & Co-Founder at OpenCore • Before that • Lars: EMEA Chief Architect at Cloudera (5+ years) • Hadoop since 2007 • Apache Committer & Apache Member • HBase (also in PMC) • Lars: O’Reilly Author: HBase – The Definitive Guide • Contact • lars.george@opencore.com • @larsgeorge Website: www.opencore.com
- 3. Agenda • Context • Data Backup Strategies • Summary
- 4. Context What do you have to look out for?
- 5. What is What? • Backup • Ability to restore data using previously taken, frozen in time data snapshots • Allows to recover deleted, or erroneously modified data • Usually backups are not current, as the most recent is not included • Disaster Recovery (DR) • Restore business and operations after a complete system failure • Includes rebuilding the environment and restoring the data from the last (good) backup • Minimize the impact on the business (financial loss)
- 6. Goals and Objectives Usually backup and DR is grounded into conditions: RTO – Recovery Time Objective • Time to recover a service • The hotter backup data is kept, the shorter the RTO • At scale, the RTO is foremost a factor of infrastructure RPO – Recovery Point Objective • Measures how much data is lost in case of a disastrous failure • The more often data is backed up, the shorter the RPO The RPO and RTO are driving cost factors and are multiplied by each other
- 7. Many Systems • Hadoop is a platform of many distributed systems • Simple tools only cover simple topics • Every system has data and/or meta data • Amount of data ranges from a few terabytes to multiple petabytes in practice • A cluster contains few to hundreds of servers What do you back up, how often, and how?
- 8. 2006 2008 2009 2010 2011 2012 2013 Core Hadoop (HDFS, MapReduce) HBase ZooKeeper Solr Pig Core Hadoop Hive Mahout HBase ZooKeeper Solr Pig Core Hadoop Sqoop Avro Hive Mahout HBase ZooKeeper Solr Pig Core Hadoop Flume Bigtop Oozie HCatalog Hue Sqoop Avro Hive Mahout HBase ZooKeeper Solr Pig YARN Core Hadoop Spark Tez Impala Kafka Drill Flume Bigtop Oozie HCatalog Hue Sqoop Avro Hive Mahout HBase ZooKeeper Solr Pig YARN Core Hadoop Parquet Sentry Spark Tez Impala Kafka Drill Flume Bigtop Oozie HCatalog Hue Sqoop Avro Hive Mahout HBase ZooKeeper Solr Pig YARN Core Hadoop The stack evolves and grows continuously! 2007 Solr Pig Core Hadoop Knox Flink Parquet Sentry Spark Tez Impala Kafka Drill Flume Bigtop Oozie HCatalog Hue Sqoop Avro Hive Mahout HBase ZooKeeper Solr Pig YARN Core Hadoop 2014 2015 Kudu RecordService Ibis Falcon Knox Flink Parquet Sentry Spark Tez Impala Kafka Drill Flume Bigtop Oozie HCatalog Hue Sqoop Avro Hive Mahout HBase ZooKeeper Solr Pig YARN Core Hadoop Evolution of the Hadoop Platform
- 9. Why is backing up data difficult? • Data at scale is difficult to move around! • You cannot cheat physics • The sheer inertia of data requires new approaches • Do not or only minimally move data as necessary • If duplicated data, use it for other purposes as well? • Multiple clusters with different workloads (Random Access vs. Analytics) • Traditional backup tools often require standardized APIs • Hadoop does not supply those necessarily, or they are inefficient here • Included backup tools in Hadoop are often rudimentary • Not all scenarios are covered, or are only partially covered
- 10. Failure Scenarios • Node Degradation • One or more nodes are slowing down or produce an increasing number of errors (and with it fewer results) – coined “The John Wayne” • Mayb cause byzantine errors, which are difficult to identify Reasons: Failures or bugs in disks, NICs, device drivers, software Hadoop can handle many such errors, but not all • Partial Node Failure • Single (redundant) components are failing completely • Example: A disk stops working • Operators can swap component at runtime Hadoop is built to handle failures like this Impact is restricted to the share of component on total capacity
- 11. Failure Scenarios (cont.) • Node Failure • Assumes preparation, like enabling HA everywhere or configure „Rack Awareness“ Reasons: Power or network outage Hadoop can handle this just fine • Network Partitioning • The cluster is split into two or more parts at random points • Causes the so-called „split brain“ problem, where each now autonomous part has to decide if it must fail, or can continue to serve request • Applications need to switch to one of the working parts of the cluster Hadoop has some support for that, but there are external dependencies What happens when the parts join the cluster again?
- 12. Failure Scenarios (cont.) • Loss of an entire data center • Complete loss of a data copy • Either switch to a warm/hot standby cluster (blue-green deployment) • Or, rebuild cluster and restore data Reasons: Power or network outage Has to be done outside of Hadoop
- 13. Data Sources • Not all Hadoop components have persistent data (or metadata) • Transient data can (should) be recomputed as needed • The number of used Hadoop components varies a lot • „Onboarding“ checklist can help to capture that • Given a set of requirements the RTO and RPO can be different • Question: How long does re-computing derived data take? • Basic Rule: The more you have, the more costly and time consuming it is • You can always omit parts, as long as everyone is OK with it (for realz!) • Cost can be capped – but not without consequence (higher RTO)
- 14. Databases in Hadoop • Many components use databases to store their state and metadata for persistency • The selection of RDBMS may have a substantial impact on that functionality Never use the ”developer option” (e.g. Derby)! The RDBMS should be highly available (HA) • Databases should be backed up and archived on a regular basis • But the question often remains: Is this a task of the Hadoop team or the (often central) IT department? • This also applies to other, external Hadoop stack systems (e.g. Storm) If possible, delegate to experienced IT team, outside of Hadoop
- 15. Data Types There are two main types of data: persisted data and metadata There is also transient data • Data concerns all user data, stored in HDFS, HBase, Solr, and so on • Can be accessed using an interface • Metadata are auxiliary information, helping to make sense of or being to access the user data • Hive Schemas • Cluster Information • Transient data often is stored in temporary files, logs, or streams
- 16. Data Consistency • An often missed (or ignored?) topic, describing what actually is inside a backup • Is the contained data consistent in itself? • Some components (NoSQL, including HDFS) cannot mark data across system boundaries in a reliable and predictable manner • Snapshots may also be of no help as they are taken asynchronously • Per regions server in HBase • Open blocks are added in HDFS • Move the task towards the application • Which application was design to do that? • When restoring data, gaps or bulges can form! • Question is: Who is responsible to handle that? • You could be tempted to add transactions...
- 17. Onboarding Checklist • Ask what is needed • How much data? • How long is retention? • Where is the data? • How often? • Define clear boundaries • What is RTO and RPO? Have user confirm and sign off explicitly!
- 18. Backup Approaches • Replication • Copy of data and modifications of one cluster to another • Some components in Hadoop support this (partially?) • HBase in near real-time, while HDFS as batch job (distcp tool) • For HDFS: Basically like the venerable rsync problem • What do you do with deleted data? How to bootstrap process? • Snapshots • Few tools have a built-in snapshot feature • HDFS and HBase • Special access to frozen-in-time data • Using special paths or system tools • Data is local and needs to be moved • How do you do this incrementally?
- 19. Backup Approaches (cont.) • Classic Backup • Store of data to a cold media • Not supplied with Hadoop • A few tools have system tools • But… Versioned? Complete? Consistent? • HA and Rack-Awareness • Does neither cover backup nor DR • Unless calling the HDFS trash functionality a backup... NOPE! • Only valid within the cluster, within the same data center
- 20. Backup Validation • After taking a backup, its integrity needs to be checked • Should consistency also be verified? • HDFS has typical checks like CRCs • Database could be restored and checked • Special test scripts? • Applications should ideally supply their own verification tools or rule sets • Make this part of the software engineering task • Use Jenkins CI as a backup und restore pipeline?
- 21. So far… • Backup is a combination of already available techniques, or a special implementation for systems that have no native support • Snapshots alone only offer local versioning • Replication is either a hot mirror, or a set of raw data structures that do not allow an instantaneous restoration • Consistency has to be handled on the application side • The required RTO und RPO is crucial for how cluster environments have to be built, and should be considered from the get go • RTO and RPO varies based on source and chosen backup strategy! • There does not seem to be a complete solution, requiring special implementations
- 22. Backup Architectures Practical scenarios (there are many more!)
- 23. Architecture #1 – Export Data Export Cost Latency Performance RTO RPO Concept • Application writes into and reads from a single cluster • Export of data to a dedicated storage service • Cheap storage arrays • Cloud storage systems (e.g. AWS S3) • Scheduled to run as a batch job on a regular basis Strength Weakness + Known architecture - Commonly slow (throttled WAN speed) + Can handle any data type (data & metadata) - Data (possibly) inaccessible unless restored first + Cost effective - High RTP and RPO Cluster A Export StorageAnwendungAnwendungApplication 💵
- 24. Architecture #2 – Replication Data Replication Cost Latency Performance RTO RPO Concept • Application writes into and reads from a single cluster • Replication of data to a standby cluster • (Possibly) smaller backup cluster with more storage and fewer CPUs • Dependent on source can run constantly or as a batch job on a regular basis Strength Weakness + Use of built-in replication (where available) - Can handle only some data types + Data accessible on backup cluster - Smaller backup cluster cannot handle all workloads + Performance a factor of parallelization - RTO and RPO depend on source Cluster A ReplicationAnwendungAnwendungApplication 💵 💵 Cluster B
- 25. Architecture #3 – Fan-out Writes Fan-Out Writes Cost Latency Performance RTO RPO Concept • Application writes into and reads from two (or more) clusters at the same time • Clusters are of same size and capacity, fan-out handled by application • Could use tools like Kafka, combined with customer (or commercial) middle-ware • ACK requires for both clusters to confirm the write • Consistency could be controlled by application (see Google Spanner and TrueTime) Strength Weakness + Clusters are independent and active-active - Highest cost + Lowest RTO and RPO - Complexity on application level + Application has full control - Validation is difficult + Can be enhanced using other tools 💵 💵 💵 Cluster A AnwendungAnwendungApplication Cluster B
- 26. Impact on Business • The basic scenarios are quite the opposites when it comes to RTO and RPO • Cost varies greatly, with #3 requiring two (or more) same size clusters In practice, any of these scenarios can be seen RTO RPO HighLow Low High 1 2 3
- 27. Summary Where to go from here?
- 28. Backup Implementation • Oozie Workflows • Main workflow that branches into sub-workflows dependent on types • Dedicated sub-workflow for each possible source • RDBMS, HBase, HDFS, Ambari/CM API, etc. • Configuration through properties files • Parameterize everything to reuse flows • Use settings to branch inside the flows • Initially create timestamp and format output directory name per run • Can be scheduled as needed
- 29. Summary Backup and DR must be part of planning and procurement from the start Many systems handle data differently, requiring special treatment Data backup and restoration has to be handled by the applications Commercial offerings are few and not fully featured
Editor's Notes
- The rapid expansion of the Hadoop ecosystem is further evidence of its meteoric adoption.