In-Memory Database
Is it really faster?
Equnix Webinar TechTalk 2021 - June 2021
MYTHBUSTERS
Copyright © 2007-2021 PT Equnix Business Solutions. Copyright is protected by law.
This document is owned by PT Equnix Business Solutions. This document contains confidential information, which
is protected by law. No part of this publication could be photocopied, reproduced or translated into another language
without permitted or the express written consent of PT Equnix Business Solutions.
Data and information regarding the proposal and its offer is for limited use and are not disclosed. The information
contained in this document is subject to change at any time without prior notice.
Copyright Statement
Our Client and Partners
What is In-Memory Database?
❏ It runs on the memory only,
never store data in the
physical disk/storage
❏ It is not persistent.
❏ Data is not really dynamic,
since the update in in batch.
❏
❏ It runs on the memory and
store data in the physical
disk/storage also.
❏ It is Persistent
❏ Data is also dynamic, there
are synchronization in
almost realtime.
How PostgreSQL actually run your Query
● In Parser, Query is parsed then raw parse tree (nodes)
is generated and analyzed into a query tree, if there are
some query can be rewrite, transform query tree using
rule system into more efficient form (ex: OR rewritten
into UNION).
● Planner will estimate the cheapest path for the
execution.
● Executor, the one really executes a user query into a
real tuple result (if expected).
What is In-Memory Database?
What is Memory Pool?
➔ A Working Memory for the
Executor/Planner database to
sort and filter page data into a
result.
➔ A Share Cache from Postgres
processes,
Memory for your OS to work,
and Kernel Cache for File
System
Memory for your PostgreSQL
Instance and Heap (Process
local memory)
Memory for Shared Buffer,
Memory Pool
How to Tune PostgreSQL into
In-Memory Performance
★ Fsync Off
★ Use RAMDISK and drop indexes
★ Or Use NVMe Disk for PG Wal
Let’s have a Demo
Thank You
Equnix Business Solutions, Pte Ltd.
8 Eu Tong Sen Street, #14-94
THE CENTRAL - Singapore 059818
Equnix Business Solutions, PT.
Sampoerna Strategic Square 25th Floor
Jl. Jend Sudirman - Jakarta 12930
Ep Date Topic Speaker
3 23 June 2021 Myth-buster: Does HA Can Help You Balance Your Load? Lucky Haryadi
4 30 June 2021 Myth-buster: Build Enterprise Level Data Center is more expensive than Cloud Lucky Haryadi
5 7 Juli 2021 Equnix’s Birthday All Equnix Team
6 14 Juli 2021 Friend or Foe: Hiring DBAs or Use Vendors? Chris Travers
7 21 Juli 2021 Friend or Foe: Is Open Source is the right choice for your system? Agatha & Yudha
8 28 Juli 2021 Are you sure that your Application is Enterprise Level? Julyanto & Hani
Don’t miss our next episodes...
Our Sponsor

Webinar2021 - In-Memory Database, is it really faster-.pdf

  • 1.
    In-Memory Database Is itreally faster? Equnix Webinar TechTalk 2021 - June 2021 MYTHBUSTERS
  • 2.
    Copyright © 2007-2021PT Equnix Business Solutions. Copyright is protected by law. This document is owned by PT Equnix Business Solutions. This document contains confidential information, which is protected by law. No part of this publication could be photocopied, reproduced or translated into another language without permitted or the express written consent of PT Equnix Business Solutions. Data and information regarding the proposal and its offer is for limited use and are not disclosed. The information contained in this document is subject to change at any time without prior notice. Copyright Statement
  • 3.
  • 4.
    What is In-MemoryDatabase? ❏ It runs on the memory only, never store data in the physical disk/storage ❏ It is not persistent. ❏ Data is not really dynamic, since the update in in batch. ❏ ❏ It runs on the memory and store data in the physical disk/storage also. ❏ It is Persistent ❏ Data is also dynamic, there are synchronization in almost realtime.
  • 5.
    How PostgreSQL actuallyrun your Query ● In Parser, Query is parsed then raw parse tree (nodes) is generated and analyzed into a query tree, if there are some query can be rewrite, transform query tree using rule system into more efficient form (ex: OR rewritten into UNION). ● Planner will estimate the cheapest path for the execution. ● Executor, the one really executes a user query into a real tuple result (if expected).
  • 6.
  • 7.
    What is MemoryPool? ➔ A Working Memory for the Executor/Planner database to sort and filter page data into a result. ➔ A Share Cache from Postgres processes, Memory for your OS to work, and Kernel Cache for File System Memory for your PostgreSQL Instance and Heap (Process local memory) Memory for Shared Buffer, Memory Pool
  • 8.
    How to TunePostgreSQL into In-Memory Performance ★ Fsync Off ★ Use RAMDISK and drop indexes ★ Or Use NVMe Disk for PG Wal
  • 10.
  • 12.
    Thank You Equnix BusinessSolutions, Pte Ltd. 8 Eu Tong Sen Street, #14-94 THE CENTRAL - Singapore 059818 Equnix Business Solutions, PT. Sampoerna Strategic Square 25th Floor Jl. Jend Sudirman - Jakarta 12930 Ep Date Topic Speaker 3 23 June 2021 Myth-buster: Does HA Can Help You Balance Your Load? Lucky Haryadi 4 30 June 2021 Myth-buster: Build Enterprise Level Data Center is more expensive than Cloud Lucky Haryadi 5 7 Juli 2021 Equnix’s Birthday All Equnix Team 6 14 Juli 2021 Friend or Foe: Hiring DBAs or Use Vendors? Chris Travers 7 21 Juli 2021 Friend or Foe: Is Open Source is the right choice for your system? Agatha & Yudha 8 28 Juli 2021 Are you sure that your Application is Enterprise Level? Julyanto & Hani Don’t miss our next episodes...
  • 13.