DB Meeting - Database High Availability Using Shared Volume Service With Offloaded WritesExport this event to calendar

Wednesday, July 23, 2014 2:30 PM EDT
Speaker: Jaemyung Kim
Abstract: Hot standby techniques are widely used to implement highly available database systems. These techniques make use of two copies of the database, an active copy and a backup that is managed by the standby. Synchronization of these two database copies is the responsibility of the database systems than manage them. However, database systems are often deployed in settings in which a reliable, persistent, network-accessible storage service (such as cloud block storage services, cluster file systems, or NAS) is available. In this paper we address the following question: how can we improve hot standby techniques in settings in which the active and standby database systems have access to a common, reliable persistent storage service?
We present SHADOW systems, a novel approach to hot standby high availability. In a SHADOW system, the active and standby database systems share access to a single logical copy of the database, which resides in the persistent shared storage. SHADOW introduces write offloading, which frees the active system from the need to update the persistent database, placing that responsibilty on the standby system instead. SHADOW systems push the task of managing database replication out of the DBMS and into the underlying storage service. We have implemented SHADOW prototypes using PostgreSQL, and we present the results of a performance evaluation that shows that SHADOW systems outperform traditional synchronous hot standby replication. Because of write offloading, SHADOW systems can potentially outperform even a standalone DBMS, while providing fast failover and durability of committed updates.
Location 
DC - William G. Davis Computer Research Centre
Room 1331
200 University Avenue West

Waterloo, ON N2L 3G1
Canada

S M T W T F S
27
28
29
30
31
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
  1. 2024 (13)
    1. December (1)
    2. October (3)
    3. June (1)
    4. May (1)
    5. April (1)
    6. March (5)
    7. February (1)
  2. 2023 (13)
    1. December (2)
    2. October (1)
    3. September (2)
    4. August (2)
    5. May (3)
    6. April (1)
    7. February (1)
    8. January (1)
  3. 2022 (6)
  4. 2021 (8)
  5. 2020 (6)
  6. 2019 (27)
  7. 2018 (26)
  8. 2017 (15)
  9. 2016 (25)
  10. 2015 (19)
  11. 2014 (34)