RAMCloud Scalable High-Performance Storage Entirely in DRAM
description
Transcript of RAMCloud Scalable High-Performance Storage Entirely in DRAM
![Page 1: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/1.jpg)
1
RAMCloudScalable High-Performance Storage Entirely in DRAM
John Ousterhout, Parag Agrawal, David Erickson, Christos Kozyrakis, Jacob Leverich, David Mazières, Subhasish Mitra, Aravind Narayanan,
Guru Parulkar, Mendel Rosenblum, Stephen M. Rumble, Eric Stratmann, and Ryan Stutsman
Stanford University
Presented by Sangjin Han(many slides are stolen from https://ramcloud.stanford.edu)
![Page 2: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/2.jpg)
2
Features (or Goals)
• Low latency: 5-10 µs (not milliseconds)• High throughput: 1M operations/s• Key-value storage with 1000s of servers• No replicas• Entirely in DRAM– Disks only for backup
• Fast recovery (1-2 secs)
![Page 3: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/3.jpg)
3
Why DRAM?
Mid-1980’s 2009 ChangeDisk capacity 30 MB 500 GB 16667xMax. transfer rate 2 MB/s 100 MB/s 50xLatency (seek & rotate) 20 ms 10 ms 2x
Capacity/bandwidth(large blocks) 15 s 5000 s 333x
Capacity/bandwidth(1KB blocks) 600 s 58 days 8333x
Jim Gray’s Rule (1KB) 5 min. 30 hours 360x
Today 5-10 years# servers 1000 4000GB/server 64GB 256GBTotal capacity 64TB 1PBTotal server cost $4M $6M$/GB $60 $6
![Page 4: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/4.jpg)
create(tableId, blob) => objectId, versionread(tableId, objectId) => blob, versionwrite(tableId, objectId, blob) => versioncwrite(tableId, objectId, blob, version) => versiondelete(tableId, objectId)
June 3, 2011 RAMCloud Overview & Status Slide 4
Data Model
Tables
Identifier (64b)Version (64b)
Blob (≤1MB)
Object
(Only overwrite ifversion matches)
![Page 5: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/5.jpg)
Master
Backup
Master
Backup
Master
Backup
Master
Backup…
Appl.
Library
Appl.
Library
Appl.
Library
Appl.
Library…
DatacenterNetwork Coordinator
1000 – 10,000 Storage Servers
1000 – 100,000 Application Servers
32-64GB/server
5
Overall Architecture
![Page 6: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/6.jpg)
6
Per-Node Architecture
• One server per data• 2-3 backups for
buffered logging
![Page 7: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/7.jpg)
7
Coordinator
• Centralized server for data placement
• E.g,
• Clients obtain a map from the coordinator– And cache
Table # First Key Last Key Server
12 0 264 - 1 192.168.0.1
47 63,742 5,723,742 192.168.0.2
… … …
![Page 8: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/8.jpg)
8
Fast Recovery (First Try)RecoveryMaster
Backups
CrashedMaster
• Reconstruct data from backup logs
• Bottleneck: disk B/W• Solution: more disks
![Page 9: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/9.jpg)
9
Fast Recovery (Second Try)
• Randomly distribute log shards• Bottleneck: recovery master• Solution: no single recovery master
RecoveryMaster
~1000Backups
CrashedMaster
![Page 10: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/10.jpg)
10
Fast Recovery (Third Try)
• Temporarily distributed recovery masters• Happy?– 35 GB recovery in 1.6s using 60 nodes (SOSP 2011)
RecoveryMasters
Backups
CrashedMaster
![Page 11: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/11.jpg)
11
Typical Datacenter Latency
![Page 12: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/12.jpg)
12
RAMCloud Latency
![Page 13: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/13.jpg)
13
Publication History
• Unveiling: SOSP 2009 WIP• White paper: Operating Systems Review 2009– What you read
• (Call for) low latency: HotOS 2011• Fast recovery: SOSP 2011• Ongoing work: index, transaction, transport, …
• Visit http://ramcloud.stanford.edu
![Page 14: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/14.jpg)
14
Thoughts & Discussion
• Garbage collection?• Infiniband vs. Ethernet– HW support for transport layer?
• Killer app?• Non volatile memory?• Synchronous vs. asynchronous query• Moving data vs. Moving code• How many papers out of this project? :P
![Page 15: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/15.jpg)
15
Thanks
![Page 16: RAMCloud Scalable High-Performance Storage Entirely in DRAM](https://reader036.fdocuments.us/reader036/viewer/2022062501/56816182550346895dd1162c/html5/thumbnails/16.jpg)
16
SORRY
• My English still s^^ks• Would you repeat the question– SLOWLY and– CLEARLY
???