Illustrates how NetApp backup software reduces backup times and capacity requirements.
First point: At a basic level, NetApp backup software behaves like traditional backup software. It copies primary data to a backup target (NetApp storage).
NetApp’s backup software provides three primary benefits over traditional backup applications—performance, capacity savings, and simplicity of management. The first thing to understand is how we can help customers increase the performance and capacity savings of their backup and recovery operations. At the top level, it’s a very basic premise—we transmit only changed data, not entire changed files. Competitors like Data Domain must still receive all the files from the server, which takes time. NetApp, however, transfers only the changes, giving us a significantly faster backup and recovery time. This is done via Open Systems SnapVault and SnapVault
2. 2 Virtual Clone FCSAN Fabric Encryption DR Backup Archive Backup Integrated Data Protectionon a single platform Satisfy multiple RTO & RPO needs Continuous availability Disaster recovery Backup and recovery Archive Secure access and compliance Simplify deployments Support physical servers, virtualization and ITaaS with the same platform Activate and configure services in minutes Integrate tightly with applications and virtual infrastructures ROBO DC Continuous Availability OFFSITE Snapshot Copies Application Integration 2
12. Gartner: Backup Trends 2010 Trend 2006 2010 Primary backup method: D2D2T ~ 60% Primary backup method: D2T 59% 25% Preferred disk target VTL CIFS, NFS, OST Average backup retention Years 90 days 6
13. Mr. Backup on… Backup! http://www.backupcentral.com/content/view/299/47 It is possible to have a complete backup system based solely on snapshots and replications You never do “restores”. In many cases, you can just start using the “backup” immediately while your primary is being repaired. Your data is never put into any magical, proprietary format. There’s no backup catalog to worry about, back up, or recover in a DR scenario. And I can’t come up with any valid objection against it. This is not only an acceptable method of backup and recovery, it’s actually a pretty dang good one. 7 Curtis Preston is an expert in backup & recovery systems and has been a consultant in the space since 1996. He's written three books on the subject, Backup & Recovery, Using SANs and NAS, and Unix Backup & Recovery. He is a frequent speaker and writer for Tech Target and other technical publications.
14. 8 Data ProtectionRequirements Solid reliability Backup in minutes Recover in minutes Simple to deploy and manage Lower TCO
15. 9 Integrated Data Protection Manage and Monitor Virtual Clone FCSAN Fabric Encryption DR Backup Archive Backup DC BU ROBO Continuous Availability Snapshot Copies Application Integration Tape
16. 10 Efficiency: Key Components Snapshot™ CopiesPoint-in-time copies that write only changed blocks. No performance penalty. Saveover80% RAID 6 Protection Protects against double disk failure with no performance penalty. Saveup to95% Saveup to95% Thin Provisioning Flexible volumes appear to be a certain size but are really a much smaller pool. DeduplicationRemoves data redundancies in primary and secondary storage. Saveup to46% Saveover80% Virtual ClonesNear-zero space, instant “virtual” copies. Only subsequent changes in cloned data set get stored. Saveup to33% Thin Replication Disk-to-disk data protection saves up to 95%
26. Day 90: 34TB on disk; 250TB on tape.SnapVault® Tape 76.8TB 34TB 19TB 12TB 12TB 13.5TB 19TB Day 1 Day 7 Day 30 Day 90 13 13
27. Customer Response to Mr. Backuphttp://blogs.netapp.com/exposed/2010/02/as-one-can-expect-from-curtis-very-through-overview-based-on-first-hand-experience-theres-not-much-to-argue-with--httpww.html 14
28. Healthcare Customer: Exchange 15 8000 Mailboxes 4.5TB Exchange DB 2TB Exchange logs EXCH EXCH EXCH Snapvault ~0,2% per day 30 daily 4 each day 90 daily 40 weekly Space used for Snapshots (Backup) 15% 975GB 35% 2.2TB
29. Financial Customer: CIFS 16 home- and group directories for 30k users 100TB primary data Space savings with Dedupe 28% Snapvault ~2% per week 14 hourly 12 daily 4 weekly 4 weekly 10-20 monthly Space used for Snapshots (Backup) 13% 13TB 45% 45TB
30. Financial Customer: OSSV Windows 17 612 Windows server WAN connected (~200 by 32kbps) 9,3TB primary data Backup gespiegelt in zwei RZ Snapvault ~2% per day 30 daily 27 weekly Space used for Snapshots (Backup) 107% 10TB
31. Financial Customer: VMware 18 9 ESX Servers 60 Server / 30 Wks (VDI) VMs 2TB primary data Space savings with dedupe 65% Space savings with dedupe 65% Snapvault ~3% per day 7 daily 100 daily Space used for Snapshots (Backup) 30% 600GB 319% 6.4TB
32. Rules of ThumbSpace used by Snapshots *this are numbers measured at our customers. Those number may vary and are solely based on the block-level change (consider e.g. database imports and re-orgs, defrag jobs, archive logs, swap files) that is generated by the application. Check our best practices on www.netapp.com/library/tr.
33. 20 Swiss Customers: Snapshots How much space is used with 424’963 copies of data? Answer: 1’084 TB (6.7%) How much space is logically represented? Answer: 265’435 TB (1600%) *Analysis over a subset of systems in Switzerland
34. NetApp provides more storage per tile Overall Systems in Switzerland (analyzed subset): 16x effective storage capacity using NetApp SnapShots
36. Rules of ThumbSpace saved by dedupe *this are numbers measured at our customers. Those number may vary and are solely based on the block-level change (consider e.g. database imports and re-orgs, defrag jobs, archive logs, swap files) that is generated by the application. Check our best practices on www.netapp.com/library/tr.
37. Dedupe in Switzerland Analysis over 3364 dedupe enabled volumes with over 1,5PB of data Average Space savings overall: 37% Average Space savings on VMware Vols: 74% Enabled on over 395PB on 30’000 systems worldwide! 24
38. 25 Deduplication for BackupNetApp Customer Data Percentage Savings Percentage Savings Data collected from NetApp AutoSupport profiling actual data Basic savings calculations can be made at secalc.com Pro forma savings estimation pool (SSET) available from Sales Engineers 25
39. Total Cost Index Acquisition, operations, support, and media costs normalized to overall total cost of NetApp® SnapVault® disk-to-disk backup Oliver Wyman TCO StudySnapVault + dedupe: 63% less expensive than tape-based backup 63% Cost Savings Administrative overhead reduction Tape acquisition and storage cost reduction 10x increase in restore speed downtime cost reduction Dedupe + block incrementals speed, reliability, consistency, efficiency * Source: Oliver Wyman Total Cost Comparison Interviews, 2006 – 2008; standardized environment with 30TB backup data + 4 remote sites 26 26
40. 27 NetApp D2D Backup 98% reduction in backup time: 16 hours to 15 minutes $100k savings for remote backups 500 person-hours saved per year on operations 3100 servers consolidated to 134 with VMware Re-architected backup Reduced full backup time from 96 hours to 30 minutes Eliminated tape loaders from backup infrastructure Recycled 225 tons of tape libraries & gear 397,000 customers, 35TB daily backups Improve backup reliability from 80% to 100% Backup 90% of data, up from 60-70% No licensed backup clients at the VM OS level 10 minute full VM restores “Using NetApp SnapVault to back up our VMware environment is saving us money. Experts who have worked on our system say our solution is better than any they’ve seen before. It’s cheaper – and delivers faster backup and recovery.” Chris Rima, Supervisor, Infrastructure Systems, Tucson Electric Power
41. 28 NetApp D2D Backup 16x faster backups, 6x faster restores 80% storage space savings Eliminated new tape infrastructure spending Budget re-allocated to SnapMirror tapeless DR 1PB backed up across 80 field offices, 6 design centers, 5000 employees 100% successful backups & restores Eliminated all future tape library purchases $500,000 in administrative savings (2 FTEs) 100% system uptime with five nines availability 70% tapeless storage environment Reduced backup times from 2 days to 30 minutes Backup failure rate reduced from 25% to <1% “If the data is online, NetApp OSSV gets it backed up. That kind of reliability is priceless. And since OSSV uses incremental updates, jobs complete in 30 minutes instead of two days. It’s rock solid.” Karl Konnerth, Director of Enterprise Operations, Affymetrix, 9/08
50. 31 Simplicity: Protection Manager Failure Readiness Protection Status Logs SnapMirror / SnapVault Events Unprotected Data Efficiency
51. Annual Benefits Where the Benefits Come From Application Layer - $57,280 annual benefits 20% reduction in end user downtime Time to recover reduced from 24 to 2 hours Database Layer - $55,680 annual benefits 30% reduction in end user downtime Time to recover reduced from 24 to 2 hours Server Layer – $40,980 annual benefits 60% increase in time to restore 10% improvement in unused storage capacity Storage Layer – $96,720 annual benefits 40% efficiency improvement in managing scripts 40% improvement in managing data protection 20% improvement in new storage utilization spend
52. NetApp D2D Backup Value Thin Replication NetApp Deduplication Open Systems SnapVault Protection Manager SnapVault SnapMirror Faster Backup in minutes More Reliable Consistently achieve 100% success rates Easier to Manage Autodiscover, autoprovision, autoprotect Network Efficient Thin Replication Storage Efficient 90% space savings w/ deduplication Cost Effective 63% less expensive than tape backups 33 33
54. SnapVault What is it? Backup license NetApp® storage via Data ONTAP® What does it do? Back up NetApp storage to NetApp storage Creates backups in seconds Block-level incrementals Centralizes ROBO backups over WAN; eliminates remote site tapes Keeps multiple backups online and available Enables self-service restore 35 35 SnapVault
55. Open Systems SnapVault What is it? Backup agent Windows®, Linux®, UNIX®, VMware® ESX What does it do? Back up any storage to NetApp® storage Block-level incrementals (only changed blocks are transferred/stored) Reduce network traffic and storage footprint Centralize ROBO backups over WAN Eliminate remote site tapes Protect data center file systems and virtual servers 36 36 Open Systems SnapVault
63. 38 38 SnapVault and SnapMirror SnapMirror® Disaster recovery Destination can be read-write Can mirror every minute RPO usually minutes True mirroring technology Deletions reflected in mirror Low RTO Supports failover/failback Bi-directional transfers IP or FC connections SnapVault® Backup and Restore Destination always read-only Can backup every hour RPO is 1 hour or more Long-term retention Keeps multiple file versions Many recovery points online CIFS/NFS accessible One-way transfers* IP connections 38
64. SnapManager What is it? Backup agent / Management agent Exchange, Oracle, SQL What does it do? Consistent Backups, Automated Management for NetApp Storage Automated, application consistent backup With SnapMirror, SnapVault, Protection Manager ‘Single button’ failover and failback Application and database recovery in minutes Large scale deployments and migrations 39 39 SnapManager
65. Protection Manager What is it? Management application SnapMirror®, SnapManager®, SnapVault®, Open Systems SnapVault What does it do? Simple policy-based management Automatically discovers new data Automatically provisions secondary storage Automatically executes data protection policies Eliminates manual configuration Provides automated exception reporting Streamlines management for up to tens of thousands of relationships 40 40 Protection Manager “We’ve grown our disk capacity by some 700% and yet have actually reduced ourdepartment size by more than 30%.” –Bradley Lauritsen, Apache Corporation team lead
66. 41 Our Credentials > 500,000 Terabytes protected for disaster recovery or backup purposes > 45,000 Deduplication-enabled systems > 65,000 Systems use replication-based data protection > 60,000 Servers backed up using our open systems backup technology > 60% F100 companies use NetApp replication-based data protection
67. 42 Strategic Direction Convergence of disk-based disaster recovery and backup Transparent protection for scale-out architectures Advanced protection for fully virtualized, cloud-based deployments
69. Unified Architecture High-End SAN Low/Midrange SAN NAS Backup Virtualization Disaster Recovery Archive & Compliance NetApp® Industry Approach FAS family Data ONTAP® FlexibilityEfficiency Agility Value Different hardware Different software Different replication Different processes Same hardware Same software Same replication Same processes 44 NetApp Confidential - Internal Use Only
70. Unified Architecture High-end SAN ReplicationProduct A Platform 1 OS1 Low-end SAN ReplicationProduct B Platform 2 OS2 ReplicationProduct C Platform 3 OS3 NAS ReplicationProduct D Platform 4 OS4 Compliance Competitors NetApp Single platform SAN & NAS Integrated Backup Any-to-any Less training NetApp SecondaryStorage 45
Editor's Notes
Make this slide a three column layoutTrend 2010 2006
Curtis says in this same blog entry that this is not a pro-NetApp blog, it’s an objective evaluation of snapshot and replication technology, including a full rundown of potential pitfalls. It just so happens that Snapshots and Replication are right in NetApp’s wheelhouse.
Primary: 2x 6070 Cluster + 1x 3050 ClusterSecondary: 1x 6070 completely on SATADR: 2x 6070 and 1x 3050 single each
Secondary: 4x 3170 Snapmirrored against each otherCa. 900 SV Relations
Primary: 3040 Metrocluster with SyncMirrorSecondary: Single 2050
Transcript:So what I then tried to do was come up with some rules of thumb, again, to share with the SEs, so that you've got something that you can put down on the table in that first meeting and say, I'd expect to see this sort of thing. We can do more analysis, we can gather more data and look at it a lot more scientifically, but if you want a rule of some, if I've got general office type data I'd expect these change rates, a daily change rate between 0.1 and 1%, weekly 1 to 7%, monthly 5 to 20. And I would expect most customers to be towards the lower end of those numbers. Like I said a few times I think, you caveat all of this with your mileage may vary. But you're probably going to get fairly good mileage. Applications can be a challenge and I found a year ago or a little over a year ago, we tended to say exchange, 10 to 20% daily change rate with our PS guys telling us that. We had some internal documents telling us that. What I found was, of the data set I looked at, and I had one customer keeping them for six week backups for a number of large exchange stores, one of the major British telcos, and their weekly changes were between 7 and 14%. One of the professional services guys in the UK had looked at about 10 different customers' exchange stores, and their daily changes averaged under 6%. What I suspect is that if you only keep a weekly one and not the daily ones in between, I suspect that that's going to be a broadly similar rate to a daily one. I'm not sure that I know enough about the internals of Exchange to know why that would be correct. So a full backup of Exchange, I would be happy saying should be under 6%. If it's a competitive situation, then a lot of them were under 2% of the ones our PS guy looked at. So you can start to spin things how you need to in the pre-sale situation with caveats that will help in the post-sale situation. NetApp has gone a long and varied history of being suicidally honest with customers and prospects and whilst I like us to take the high road and be the customer's kind of a trusted adviser, I also like us to win the business. Databases, again, could vary enormously, but as a rule of thumb, as a starting point I'd suggest maybe a 1% change rate. And this, as I say at the top, detailed analysis is not possible. So when you're in those first meetings, just saying the sort of things you can get, these are the sorts of savings you'd expect. And this is without any deduplication processing overhead.Author’s Original Notes:I expect the majority of customers to lean to the lower end of these estimates.
Transcript:So what I then tried to do was come up with some rules of thumb, again, to share with the SEs, so that you've got something that you can put down on the table in that first meeting and say, I'd expect to see this sort of thing. We can do more analysis, we can gather more data and look at it a lot more scientifically, but if you want a rule of some, if I've got general office type data I'd expect these change rates, a daily change rate between 0.1 and 1%, weekly 1 to 7%, monthly 5 to 20. And I would expect most customers to be towards the lower end of those numbers. Like I said a few times I think, you caveat all of this with your mileage may vary. But you're probably going to get fairly good mileage. Applications can be a challenge and I found a year ago or a little over a year ago, we tended to say exchange, 10 to 20% daily change rate with our PS guys telling us that. We had some internal documents telling us that. What I found was, of the data set I looked at, and I had one customer keeping them for six week backups for a number of large exchange stores, one of the major British telcos, and their weekly changes were between 7 and 14%. One of the professional services guys in the UK had looked at about 10 different customers' exchange stores, and their daily changes averaged under 6%. What I suspect is that if you only keep a weekly one and not the daily ones in between, I suspect that that's going to be a broadly similar rate to a daily one. I'm not sure that I know enough about the internals of Exchange to know why that would be correct. So a full backup of Exchange, I would be happy saying should be under 6%. If it's a competitive situation, then a lot of them were under 2% of the ones our PS guy looked at. So you can start to spin things how you need to in the pre-sale situation with caveats that will help in the post-sale situation. NetApp has gone a long and varied history of being suicidally honest with customers and prospects and whilst I like us to take the high road and be the customer's kind of a trusted adviser, I also like us to win the business. Databases, again, could vary enormously, but as a rule of thumb, as a starting point I'd suggest maybe a 1% change rate. And this, as I say at the top, detailed analysis is not possible. So when you're in those first meetings, just saying the sort of things you can get, these are the sorts of savings you'd expect. And this is without any deduplication processing overhead.Author’s Original Notes:I expect the majority of customers to lean to the lower end of these estimates.