Rbd - ordered crash-consistent write-back caching extension » History » Version 1
Tushar Gohad, 12/02/2015 06:22 AM
1 | 1 | Tushar Gohad | h1. Rbd - crash-consistent ordered write-back caching extension |
---|---|---|---|
2 | |||
3 | h3. Rationale |
||
4 | |||
5 | RBD cache currently supports object-based DRAM caching only, with no ordered write-back support. Content cacheability is limited by the size of DRAM - proposal here is to extend librbd to support - |
||
6 | |||
7 | * A new librbd read cache to support LBA-based caching with DRAM/non-volatile storage backends |
||
8 | * An ordered write-back cache that maintains checkpoints internally (or is structured as a data journal), such that writes that get flushed back to the cluster are always *crash consistent*. Even if one were to lose the client cache entirely, the disk image is still holding a valid file system that looks like it is just a little bit stale [1] |
||
9 | * External caching plug-in interface – kernel and usermode |
||
10 | |||
11 | This enhancement has been discussed at the 2015 Ceph hackathon and also at the Tokyo summit with the Ceph/RBD core. |
||
12 | |||
13 | This proposal is based on discussions in [1], [2] with Sage and Josh. Proof points for client-side caching in [2], [3] and external caching plugin example in Maciej’s CAS POC presentation at the Ceph Performance Weekly Meeting on Nov 25, 2015. |
||
14 | |||
15 | Tunable ordered Write-back caching option should provide boost for streaming, log storage, some VDI workloads and eventually consistent databases. Where write-back is not suitable, a read-only (write-through) option should help improve access latencies - VDI and database workloads for example. |
||
16 | |||
17 | h3. Owners |
||
18 | |||
19 | * Anjaneya Chagam (Intel) |
||
20 | * Tushar Gohad (Intel) |
||
21 | * Stephen Blinick (Intel) |
||
22 | |||
23 | h3. Interested Parties |
||
24 | |||
25 | * Sage Weil, Josh Durgin |
||
26 | |||
27 | h3. Project Phases (current proposal) |
||
28 | |||
29 | |||
30 | * Phase1: private, LBA-level, read-only cache - plumb in new generic caching layer in librbd for non-volatile storage-backend – possible ObjectCacher replacement. Read-only cache, designed with scope for write-back caching and pluggability extensions in later phases |
||
31 | * Phase2: node-local, shared read-only cache. Private ordered/crash-consistent write-back cache |
||
32 | * Phase3: node-local, shared ordered/crash-consistent write-back cache. External caching plugins – Intel CAS, dm-cache, etc |
||
33 | |||
34 | |||
35 | h3. References: |
||
36 | |||
37 | [1] http://ceph-users.ceph.narkive.com/0LDAkjYy/persistent-write-back-cache |
||
38 | [2] Ceph Hackathon’15 and Tokyo Summit Discussions on RBD write-back caching |
||
39 | [3] Whitepaper: database workloads using RBD with dm-cache and Ceph cache tiering |