SAP HANA Operation Expert Summit BUILD - High Availability & Disaster Recovery Dr. Ralf Czekalla/SAP HANA Product Management May 09, 2014 Customer
Disclaimer This presentation outlines our general product direction and should not be relied on in making a purchase decision. This presentation is not subject to your license agreement or any other agreement with SAP. SAP has no obligation to pursue any course of business outlined in this presentation or to develop or release any functionality mentioned in this presentation. This presentation and SAP's strategy and possible future developments are subject to change and may be changed by SAP at any time for any reason without notice. This document is provided without a warranty of any kind, either express or implied, including but not limited to, the implied warranties of merchantability, fitness for a particular purpose, or noninfringement. SAP assumes no responsibility for errors or omissions in this document, except if such damages were caused by SAP intentionally or grossly negligent. 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 2
High Availability Disaster Recovery Agenda Business Continuity High Availability Disaster recovery per Center between Centers 1 SAP HANA Host Auto-Failover (Scale-Out with Standby) 3 SAP HANA Storage Replication 2 SAP HANA System Replication Performance Optimized 4 SAP HANA System Replication Performance Optimized Cost Optimized Cost Optimized 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 3
HA & DR Concepts in general Classification of solutions system operational operation resumed Sync or backup RPO RTO design & prepare detect recover perf. ramp time KPIs: Recovery Point Objective (RPO) = worst-case data-loss Recovery Time Objective (RTO) = time to recover from outage Solution Used for Cost RPO RTO Perf. ramp Backup & Recovery HA & DR $ high high med SAP HANA Host Auto-Failover HA $ 0 med long SAP HANA Storage Replication w/ QA, Dev. DR $$ 0* med long SAP HANA System Replication HA & DR $$$ 0* low short SAP HANA System Replication w/ QA, Dev. HA & DR $**/$$ 0* med long ** single host installations *synchronous solution 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 4
High Availability Options 1 Scale-Out or Host Auto-Failover
Storage Connector API Shared Storage SAN Storage High Availability for SAP HANA: Host Auto-Failover High Availability configuration N active s in one cluster M standby (s) in one cluster Shared file system for all s Services and index on all nodes Statistics (only on one active ) active on Standby Failover X fails N+1 reads indexes from shared storage and connects to logical connection of X Storage Connector API ensures remount of necessary disk areas (Note 1900823 - Storage Connector API Attachments) 1 2 3 4 5 6 Standby 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 6
High Availability for SAP HANA Host Auto-Failover (standby) Different implementation of High Availability by HW partners Using storage solution inside Using internal disk Standby GPFS GPFS 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 7
High Availability Options 2 SAP HANA System Replication
HA Solution Partner HA Solution Partner High Availability for SAP HANA: System Replication Performance Optimized Center 1 OS: DNS, hostnames, virt. IPs Internal Primary (active) Clients Transfer by SAP HANA database kernel Application s Secondary (active, data pre-loaded) Internal Performance optimized option Secondary system completely used for the preparation of a possible take-over Resources used for data pre-load on Secondary Take-overs and Performance Ramp shortened maximally 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 9
HA Solution Partner HA Solution Partner High Availability for SAP HANA: System Replication Cost Optimized Center 1 OS: DNS, hostnames, virt. IPs Internal Primary (active) Clients Transfer by SAP HANA database kernel Application s Secondary PRD shadow operation PRD QA/DEV running Internal QA/DEV Cost optimized with Operating non-prod systems on Secondary Resources freed (no data pre-load) to be offered to one or more non-prod installations During take-over the non-prod operation has to be ended Take-over performance similar to cold start-up 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 10
High Availability for SAP HANA Comparison of minimal setups for available solutions Scale-Out 1+1 (Host Auto-Failover) active Master standby System Replication with both alternatives (1+1) Internal Primary (active) Transfer by SAP HANA database kernel Secondary (active, data pre-loaded) Internal Performance optimized (data pre-loaded) or Cost optimized Secondary PRD shadow operation QA/DEV running Storage System PRD QA/DEV 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 11
Disaster Recovery Options 3 SAP HANA Storage Replication
Storage Mirroring HA Solution Partner HA Solution Partner Disaster Recovery for SAP HANA: Storage Replication Cluster across Centers with non-prod on 2 nd site Center 1 OS: DNS, hostnames Primary Clients Application s Center 2 Secondary Prod. (inactive), QA&DEV (active) Arrangement usually offered with a strong part of hardware partners involvement Support issues handled by/routed to HW partners TCO reduction by combined operation with non-prod on Secondary OS: Mounts Needs another disk stack for non-prod usage load Cluster management often included and delivered as a whole package 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 13
HA & DR solutions from hardware vendors with certified HANA offering May 2014* (only China) Scale Out (BW) S - L M M, L M M S - L M M M High Availability X X X X X X X X X DR Storage Repl.: Async DR Storage Repl.: Sync X X X planned X X X X X X Another asynchronous solutions for SAP HANA Storage Replication certified, SAP note 1755396 adapted * For most up to date list please go to the SAP Product Availability Matrix (Westmere) or http://scn.sap.com/docs/doc-52522 (IvyBridge) 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 14
Disaster Recovery Options 4 SAP HANA System Replication
HA Solution Partner HA Solution Partner Disaster Recovery for SAP HANA: System Replication Cluster across Centers with DB controlled transfer Center 1 OS: DNS, hostnames, virt. IPs Primary (active) OS: Mounts Clients Transfer by SAP HANA database kernel Application s Center 2 Secondary (active, data pre-loaded) Performance optimized option Faster Take-Over Shortened Performance Ramp (seconds to less minutes) SYNC & ASYNC possible Several cluster options Some HW Partners offer pre-packaged options Step-by-Step Implementation Guide: https://scn.sap.com/docs/doc-47702 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 16
HA Solution Partner HA Solution Partner Disaster Recovery for SAP HANA: System Replication Cluster across Centers with non-prod operation on 2 nd site Center 1 Clients Application s Center 2 Cost Optimized option SYNC & ASYNC possible OS: DNS, hostnames, virt. IPs Primary (active) OS: Mounts Transfer by SAP HANA database kernel Secondary (active,) PRD shadow operation QA/DEV running TCO reduction by combined operation with non-prod on Secondary Needs another disk stack for non-prod usage load Step-by-Step Implementation Guide: https://scn.sap.com/docs/doc-47702 PRD QA/DEV 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 17
SAP HANA in Centers Cluster Manager with direct SAP HANA System Replication Support HP Service Guard Directly available from HP and HP drives the implementation process Link for further information: o Reference Guide: HP ServiceGuard Extensions, March 2014 o User Guide: Managing HP ServiceGuard Extensions for SAP for Linux, December 2013 SUSE Cluster Included with SUSE SLES for SAP Applications Blog: Fail-Safe Operation of SAP HANA : SUSE Extends Its High-Availability Solution SAP Landscape Virtualization Manager (SAP LVM) Consulting package to create a cluster manager with SAP LVM available Other cluster managers In pipeline or can be adapted with individual consulting packages Note: No solution certification is offered for cluster managers Support is handled by solution partner directly 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 18
Further options for System Replication
SAP HANA Zero Downtime maintenance Featured by SAP NetWeaver ABAP stack As an evolution of Near Zero Downtime Maintenance, SPS7 now offers Zero Downtime Maintenance Based on connectivity suspend feature of the SAP NetWeaver ABAP stack (SAP note 1913302) o DBSL of the database interface decouples transaction management between ABAP and SAP HANA database o This keeps transaction on ABAP layer alive and allows to change components (software versions) on the layers below on secondary (shadow) SAP HANA instance Further information also in Step-by-Step Implementation Guide for SAP HANA System Replication: https://scn.sap.com/docs/doc-47702 Hardware mix (SAP note 1984882 - Using SAP HANA System Replication for Hardware Exchange with minimum Downtime) Walldorf Primary SAP NetWeaver ABAP DBSL Software Upgrade Order SAP HANA Version SAP HANA Version +1 Sync/Async mirrored redo log writing Transport incremental data Rot Secondary 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 20
Center Setups with Pre-load Option Multi Tier System Replication Cascading Systems Center Center Sync Async Production Local standby with data preload Remote standby system with or without preload (mixed usage with non-prod.) Available with SAP HANA (Three cascading systems with SPS7) 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 21
SAP Business Systems @ SAP SAP HANA Usage in Production
SAP Business Systems @ SAP SAP HANA usage in production SAP s financial accounting system running SAP Business Suite powered by SAP HANA Scale-Out 2+1 (local High Availability) Secured with SAP HANA System Replication SAP s CRM system running SAP CRM powered by SAP HANA Scale-out 2+1(local High Availability) Secured with SAP HANA System Replication Center: St. Leon-Rot Center: Walldorf SAP s BW system running SAP CRM powered by SAP HANA Scale-Out 8+1(local High Availability) No DR Distance: about 10 km Redundant connections: 3 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 23
SAP s financial accounting system running SAP Business Suite powered by SAP HANA Technical details Business SAP Application SAP Business Suite powered by SAP HANA Number of named Users 65000 Transactions per day tbc Hardware Manufacturer IBM CPU Model 8 CPUs with 10 core, Xeon E7-8870 @2.4GHz Memory 2 x 4096 GB Operating System SUSE Linux Enterprise 11.2 Scale-out Scale-out plus standby (2+1) Storage IBM GPFS file system Restart Restart Time (min)/takeover 5 15 min Performance Ramp 1 h 1,5 h Loading RowStore 3 min 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 24
SAP s financial accounting system running SAP Business Suite powered by SAP HANA base DB Version HANA 1.0 Revision 73 DB Usage Custom DR Solution SAP HANA System Replication Replication mode syncmem Distance between Centers 10 km Bandwidth for System Replication 10 Gbit Total database size Column Store size Row Store size Avg local log write wait time Avg remote log write wait time Delta data shipping throughput backup (log volume) per day Transferred data volume between data centers per day 1,7 TB 1460 GB 140 GB 210 µs (for avg. log buffer size of 9,5 MB) 448 µs (for avg. log buffer size of 9,5 MB) 108 MB/sec 165 GB 1,7 TB Backup Storage Avg. size of full data backup Throughput Avg. runtime of backup (to FS) Estimated total recovery time Estimated data recovery time Estimated log recovery time IBM GPFS file system backup (local disks) 1,7 TB 775 MB/s (2800 GB/h) 41 min 340 min 70 min (~1,7 TB) 270 min (~3,7 TB log backups of 3 weeks) [1] In a similar productive SAP system the recovery was done with a three weeks old backup (1,8 TB) and the log backups of these three weeks (7 TB). The data recovery took 70 minutes and the log recovery took 9 hours makes roughly 10 hours recovery time for this system. 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 25
Roadmap
Disaster Recovery for SAP HANA News with SAP HANA SPS07 and Beyond System Replication extension Multi Tier System Replication allowing first 1:n setups: Three cascading systems in a row Zero Downtime maintenance Compressed log transfer Basic encryption for internal communication Planned beyond System Replication extension o Pure -based transfer o Backup on shadow instance o Online Add Host & Remove Host on Secondary o Active/Active Operation (r/o reporting on Sec.) o More 1:n relationships for shadow instances o Time travel via internal snapshots on shadow instance to handle logical errors o More asymmetric options (n m) o Time delay option between sites Shipping o Based on backup files (initial data, sub sequential log, steady roll forward) 2014 SAP AG or an SAP affiliate company. All rights reserved. Customer 27
Thank you Contact information: Dr. Ralf Czekalla Product Management TIP In-Memory Platform Ralf.Czekalla@sap.com http://www.saphana.com/ 2014 SAP AG or an SAP affiliate company. All rights reserved.