Tag line tag line Flex Cache Date August

  • Slides: 12
Download presentation
Tag line, tag line Flex. Cache Date August 7, 2008 Presenter Marty Turner

Tag line, tag line Flex. Cache Date August 7, 2008 Presenter Marty Turner

Flex. Cache™ Overview ¡ Storage performance accelerator ¡ Cache storage protocol – Supports NFS

Flex. Cache™ Overview ¡ Storage performance accelerator ¡ Cache storage protocol – Supports NFS v 2/v 3 ¡ Cache architecture – Memory and disk – Volume granularity – Write-through © 2008 Net. App. All rights reserved. Origin System Origin Volume Flex. Cache Volume NFS Clients Flex. Cache System 2

Flex. Cache Overview ¡Significant value across various workloads – – Shared data sets Repeated

Flex. Cache Overview ¡Significant value across various workloads – – Shared data sets Repeated access Random reads High IOPS/concurrency ¡Across different verticals – Financial – EDA – Software build © 2008 Net. App. All rights reserved. – Movie rendering – Energy exploration – Biotech 3

Flex. Cache Use Cases Three primary: ¡ (A) Scale out performance ¡ (B) Reduce

Flex. Cache Use Cases Three primary: ¡ (A) Scale out performance ¡ (B) Reduce WAN latency ¡ (C) Offload resources Also: ¡ Tiered storage Flex. Cache Volumes Origin System Origin Volume NFS Clients Caching Systems (A) Flex. Cache Volume Caching System (C) WAN (B) NFS Clients © 2008 Net. App. All rights reserved. 4

Scale Out Performance Challenges ¡ Compute farms have huge demand NFS Clients ¡ Storage

Scale Out Performance Challenges ¡ Compute farms have huge demand NFS Clients ¡ Storage performance dictates grid growth Flex. Cache Volume Origin System Origin Volume Caching System …… Caching System S C A L E O U T Flex. Cache Volume © 2008 Net. App. All rights reserved. Flex. Cache™ increases IOPS ¡ Scales capability for shared data ¡ Balances client load across multiple caching systems Benefits ¡ Extensive scalability ¡ No need to replicate/mirror data 5

Reduce WAN Latency Challenges Origin Volume ¡ Wide area networks are slow ¡ Ensure

Reduce WAN Latency Challenges Origin Volume ¡ Wide area networks are slow ¡ Ensure nodes are always processing Flex. Cache Volume Caching System Flex. Cache™ reduces WAN latency ¡ Caches data near client ¡ Faster access times ¡ Avoids the overhead of replication WAN Origin System Benefits NFS Clients © 2008 Net. App. All rights reserved. ¡ Reduced latency for NFS clients ¡ Nondisruptive ¡ Ease of management ¡ Disk-space efficient ¡ Reduces WAN utilization 6

Offload Resources Challenge Flex. Cache Volumes Origin System NFS Clients Caching Systems ¡ Overloaded

Offload Resources Challenge Flex. Cache Volumes Origin System NFS Clients Caching Systems ¡ Overloaded storage system or network infrastructure Flex. Cache™ offloads resources Origin Volume Flex. Cache Volume Caching System ¡ Origin available for other tasks ¡ NFS ops served from cache Benefits ¡ Offloads storage ¡ Offloads network ¡ Simplifies administration WAN NFS Clients © 2008 Net. App. All rights reserved. 7

Flex. Cache as Tiered Storage Challenge Primary 1 NFS Clients (Flex. Cache) ¡ Provide

Flex. Cache as Tiered Storage Challenge Primary 1 NFS Clients (Flex. Cache) ¡ Provide cost-effective tiered storage Secondary (Origin) Cached Data Fewer Drives Data A’, B’ App Drives Migration Primary 2 NFS Clients (Flex. Cache) All Data Low Cost Drives Data A, B, C Cached Data Fewer Drives Data B’, C’ © 2008 Net. App. All rights reserved. Flex. Cache paradigm ¡ Primary is Flex. Cache ¡ Secondary is origin ¡ Migration is application driven Benefits ¡ No migration policies ¡ Sharing data is simple ¡ Data management on secondary 8

Flex. Cache Deployment Choices ¡ Flex. Cache™ license: Multipurpose or repurposed FAS – Any

Flex. Cache Deployment Choices ¡ Flex. Cache™ license: Multipurpose or repurposed FAS – Any FAS or V-Series with Data ONTAP® 7 G ¡ Net. App® SA system: Dedicated, cost-effective cache Target Use Case IOPS Base Config Memory Expandability SA 200 Entry Data Center/Remote Caching 11, 500 2 GB No memory expansion Storage up to 104 TB SA 300 Mid Data Center Caching 34, 500 8 GB Memory up to 40 GB Storage up to 504 TB SA 600 Large Data Center Caching 220, 000 32 GB Memory up to 112 GB Storage up to 1176 TB © 2008 Net. App. All rights reserved. 9

Customer Success: Cadence—Semiconductor EDA Compute Farm Caching Tiers FAS 3050 Origin V 6070 A

Customer Success: Cadence—Semiconductor EDA Compute Farm Caching Tiers FAS 3050 Origin V 6070 A FAS 3050 IBM Shark SAN FAS 3050 “The true performance of caching will help us scale to 10, 000 servers in the compute farm. ” —Cadence success story © 2008 Net. App. All rights reserved. 10

Customer Success: Major Studio—Movie Rendering Render Farm ~750 nodes 18 Caching Systems WAN 10

Customer Success: Major Studio—Movie Rendering Render Farm ~750 nodes 18 Caching Systems WAN 10 Origin Systems “We cannot render movies without Flex. Cache. ” —Customer quote © 2008 Net. App. All rights reserved. 11

Thank you! © 2008 Net. App. All rights reserved. 12

Thank you! © 2008 Net. App. All rights reserved. 12