Download presentation
Presentation is loading. Please wait.
Published byGwenda Bates Modified over 9 years ago
1
Objection Handling Module #4 15 Slides – audio run time: 60 minutes CBT
2
COPAN Systems CONFIDENTIAL 2 Major Objections Drive Life and Reliability What is Drive MTBF, Failure Rates? RAID5 versus RAID6/Dual Parity Hot-Swap Drives and Serviceability # of VTDs /# of Streams per Shelf Dual Path/Failover in B/R Application Single Instancing Disk to Disk backup vs. VTL backup CBT
3
Drive Life and Reliability CBT
4
COPAN Systems CONFIDENTIAL 4 Drive Life and Reliability Disk Reliability: POWER-MANAGED RAID and DISK AEROBICS Benefits Compared to std SATA disk, COPAN has less than ¼ the failure rate Field MTBF: more than 4X SATA disks, more than 2X FC disks Service Life: expect more than 4X Disk Reliability and TCO benefits Per 1000 drives, expect only 3 drives to fail/yr vs. 15 drives in case of competitor’s platform COPAN: 0.3% failures/yr SATA: 1.45% failures/yr Competitors have ~5X drive replacements 17 touches versus 1 touch for COPAN Data Reliability Benefits Fewer failures 1/23 Data Loss 600K hrs = 68 yrs 2.64M hrs = 331 yrs CBT
5
COPAN Systems CONFIDENTIAL 5 Points of Note for MTBF Charts 79% less failures than other std SATA competitors MTBF is calculated from total number of hours for all drives (drive-hours) per number of returned drives This is real field data – not controlled tests done by HDD vendors Drive hours increasing much faster than failure occurrences Validates significant benefits of POWER-MANAGED RAID and DISK AEROBICS software CBT
6
COPAN Systems CONFIDENTIAL 6 Objection Handling: Dual Parity/RAID6 Std SATA vendors have over 4X failure rates of COPAN Data loss probability with std SATA is almost 22X COPAN’s (two drive failures) Some vendors, e.g., NTAP, provide dual-parity (DP) option to reduce compensate their lower data reliability with Dual Parity COPAN’s Power-Managed RAID and COPAN’s Power-Managed RAID and Disk Aerobics improves data reliability higher than primary FC disk RAID arrays Disk Aerobics improves data reliability without sacrificing capacity, performance and complexity of dual-parity Dual Parity RAID6 40% overhead CBT
7
COPAN Systems CONFIDENTIAL 7 Hot-Swap and Serviceability Std SATA drive MTBF of 600,000 hrs With 112 drives, expect 1 failure every 223 days With COPAN’s POWER-MANAGED RAID, eff. MTBF 2.83M hrs With 112 Disks, expect 1 failure every 1080 days (about 3 years) Drive failure rate per 112 drives (COPAN shelf) Probability that 1 or more drives fail/yr is 29% in vs. 81% for other vendors Revolution 200 series has 32 spares/system vs 4-8 spares for other vendors (e.g., NetApp): getting down to 1 spare/yr is 0.5% Other vendors need to hot-swap drives because they have 1 spare per shelf or 2 shelves: more than 50% chance 2 drives will fail within a year Other vendors cannot wait for 2 nd drive to fail – their data loss probability is 23X COPAN’s With COPAN’s much lower failure rates – dead drives don’t have to be replaced immediately! COPAN has changed the rules of serviceability Scheduled annual service to replace failed drives Unlike other vendors, not required to immediately replace drives COPAN has less intrusive operations and lower TCO CBT
8
COPAN Systems CONFIDENTIAL 8 Summary of COPAN Reliability COPAN’s POWER-MANAGED RAID and DISK AEROBICS Provides Unparalleled Benefits Failure Rate: COPAN has less than ¼ the failure rate of competitors: Drive Service Life will be more than 4X Data Reliability: competitors’ data loss probability with std SATA is almost 23X COPAN’s Lower TCO – with reliability and spares: competitors will have 17 system touches vs 1 scheduled touch No Need for Dual Parity: COPAN has improved data reliability 6X over FC primary disk, 23X over SATA without sacrificing capacity, performance and complexity of dual-parity CBT
9
# of VTDs /# of Streams per Shelf CBT
10
COPAN Systems CONFIDENTIAL 10 Configuring Backups: #Streams/Shelf and # VTDs Concern: COPAN’s 56 VTDs/platform limits uses in some B/R scenarios versus some competitor VTLs Issues Power management limits power/shelf and VTDs/shelf Need proper configuration to balance Bandwidth vs #Streams Response: the best solution with B/R applications is not necessarily connecting each LAN client to a single VTD! B/R applications do not need to connect to very large number of VTDs: this is an old tape model when drives were slow Connecting “skinny” clients directly to COPAN VTD not recommended: highly underutilized bandwidth Major B/R apps can work with smaller number of streams NBU and Legato Networker both support multiplexing: many LAN client streams can be multiplexed to a few VTDs TSM uses a disk cache: 1000s of clients can write to a disk cache and then stream to the COPAN VTDs in the backend CBT
11
COPAN Systems CONFIDENTIAL 11 Responding to #VTDs Objection Major B/R applications are not limited by less than 100 VTDs Configurations with NetBackup should use multiplexing to utilize COPAN bandwidth Configurations with TSM should use disk cache to utilize COPAN bandwidth The only situation where we have a disadvantage is if the customer decides to directly connect many (>56) clients directly via FC to COPAN B/R Application Connectivity#VTDs : #StreamsBackup MB/s per stream Recovery MBs per stream Symantec NBU FC1 : 1 #streams=#VTDs = n 180/n≤ 90 Symantec NBU GigE1: many (m) VTD = m streams 180/(mn)≤ 90/m TSMFC or GigENo relation #streams = n Disk Cache Bandwidth/n ≤ 90 CBT
12
Dual Path/Failover in B/R Application CBT
13
COPAN Systems CONFIDENTIAL 13 Dual Path and Failover Current applications in backup recovery do not have instantaneous failover need HA not needed for second copy of data Physical tape libraries need redundant copies - managed by B/R apps VTLs provide fault tolerance for potential single points of failure Fault tolerance needed for applications that store the only copy File Archive Native MAID applications CBT
14
COPAN Systems CONFIDENTIAL 14 Planned Dual Path - Failover COPAN’s Committed Program: Q3 release of an Active-Active HA MAID platform to support all personalities Dual path from 2 Rack Controllers to 2 Shelf Controllers to Canister – to all data No single point of failure Active- Active shelf failover Failover model in 3-tier Architecture: Rack Controller (Active-Active) Shelf Controller (Active-Active) Canister Controller (Active-Passive) Points of Failure Tape LibraryVirtual Tape Library Tape cartridgeNo fault tolerance, need second copy of tape Data protection with RAID Tape Transport Device Multiple tape transport in Library Dual Rack Controller, Dual Shelf Controller, Dual-path Canister Controller Storage interconnect Not availableFibre Channel spoofing for transparent path failover CBT
15
Single Instancing CBT
16
COPAN Systems CONFIDENTIAL 16 Single Instance – How it Works Appliance decomposes files into segments, creates hashes and compares them against stored segment hashes If hashes match, most likely incoming data segment match is already stored. If not, segment is stored and its hash is added to an index (or DB) Appliance can avoid storing based on the hash match or retrieve the stored segment and verify it is a valid match If appliance determines it is not present in the storage, then segment is stored CBT
17
COPAN Systems CONFIDENTIAL 17 Single Instance Challenges Single Instance creates significant performance problems to meet backup window: < 200 MBs! Single instance creates single point of data failure There is no redundancy – loss of any data in the first instance can result in losing all 10-20 backups Data loss potential Comparison based on a single hash can result in data loss Current vendors using std SATA storage with much lower reliability than COPAN MAID TSM and Symantec NBU now support single instance features TSM “subfile backup” for local or remote backups Veritas PureDisk for remote office backup CBT
18
COPAN Systems CONFIDENTIAL 18 Objection Handling Single Instance Single Instance VTLs have created serious challenges Severe performance problems Single point of data failure Data loss potential Single Instance competitors use generic SATA arrays with very poor reliability Major backup applications (TSM and NBU) are now providing single instance features Where Single Instance is Relevant: create a low- footprint image offline that is used for remote replication Could reduce 95% bandwidth needs ‘Intelligent Remote Replication’ CBT
19
D2D Backup vs VTL Backup CBT
20
COPAN Systems CONFIDENTIAL 20 Objection Handling: D2D vs VTL Backups Some B/R environments are using D2D backup Less reliable disk platforms unlike COPAN Deployed as D2D2T with data ending up on tape: less reliable and performance than COPAN MAID Learning curve: this is new application - needs new process VTL B/R is still most well-understood D2D B/R emerging – MAID disk can enable this given the significant benefits of the platform CBT
21
COPAN Systems CONFIDENTIAL 21 D2D Backup vs. VTL Backup CBT
22
COPAN Systems provides intelligent storage solutions that unlock the value of long-term data. Thank You CBT
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.