DZone
Database Zone
Thanks for visiting DZone today,
Edit Profile
  • Manage Email Subscriptions
  • How to Post to DZone
  • Article Submission Guidelines
Sign Out View Profile
  • Post an Article
  • Manage My Drafts
Over 2 million developers have joined DZone.
Log In / Join
  • Refcardz
  • Trend Reports
  • Webinars
  • Zones
  • |
    • Agile
    • AI
    • Big Data
    • Cloud
    • Database
    • DevOps
    • Integration
    • IoT
    • Java
    • Microservices
    • Open Source
    • Performance
    • Security
    • Web Dev
DZone > Database Zone > Percona Memory Engine for MongoDB

Percona Memory Engine for MongoDB

Learn more about Percona's memory engine, designed for MongoDB. See how it compares to WiredTiger and where it runs into problems.

Denis Protyvenskyi user avatar by
Denis Protyvenskyi
·
Aug. 13, 16 · Database Zone · Analysis
Like (1)
Save
Tweet
3.96K Views

Join the DZone community and get the full member experience.

Join For Free

This post discusses Percona Server for MongoDB’s new in-memory storage engine, Percona Memory Engine for MongoDB.

Percona Server for MongoDB introduced the Memory Engine starting with the 3.2.8-2.0 version. To use it, run Percona Server for MongoDB with the --storageEngine=inMemory option.

In-memory is a special configuration of WiredTiger that doesn’t store user data on disk. With this engine, data fully resides in the virtual memory of the system (and might get lost on server shutdown).

Despite the fact that the engine is purely in-memory, it writes a small amount of diagnostic data and statistics to disk. The latter can be controlled with the --inMemoryStatisticsLogDelaySecs option. The --dbpath option controls where to store the files. Generally, in-memory cannot run on the database directory previously used by any other engine (including WiredTiger).

The engine uses the desired amount of memory when configured with the --inMemorySizeGB option. This option takes fractional numbers to allow precise memory size specification. When you reach the specified memory limit, a WT_CACHE_FULL error is returned for all kinds of operations that cause user data size to grow. These include inserting new documents, creating indexes, updating documents by adding or extending fields, running aggregation workflow, and others. However, you can still perform read queries on a full engine.

Since Percona Memory Engine executes fewer operations and makes no disk I/O system calls, it performs better compared to conventional durable storage engines, including WiredTiger’s standard disk-based configuration.

Performance

The following graphs show Percona Memory Engine versus WiredTiger performance. Both engines use the default configuration with 140GB cache size specified. The hardware is 56-core Intel(R) Xeon(R) CPU E5-2683 v3 @ 2.00GHz with 256GB of RAM and RAID1 2xHDD. Test data set is about cache size and fully fits in memory.

Memory Engine vs WiredTiger Insert


Memory Engine vs WiredTiger OLTP


You can clearly see that Percona Memory Engine has better throughput and less jitter on all kinds of workloads. Checkpointing, however, can cause jitters in WiredTiger — and are absent in Percona Memory Engine, as there’s no need to periodically sync in-memory data structures with their on-disk representations.

However, the performance of Percona Memory Engine drops when it’s about to become full (currently, when it’s 99% full). We’ve marked this issue as fixed, but it still crops up in extreme cases.

Percona Memory Engine might use up to 1.5 times more memory above the set configuration when it’s close to full. WiredTiger almost never exceeds the specified cache memory limit. This might change in future versions. But current users should avoid possible swapping or OOM-killing of the server with Percona Memory Engine if (mis)configured to use all or close to all of available system RAM.

You can download the latest version of Percona Server for MongoDB, which includes the new Percona Memory Engine feature, here.

Engine Memory (storage engine) MongoDB

Published at DZone with permission of Denis Protyvenskyi, DZone MVB. See the original article here.

Opinions expressed by DZone contributors are their own.

Popular on DZone

  • Optional in Java: When Better Code Is Not an Alternative
  • Automatically Creating Microservices Architecture Diagrams
  • Pull Request vs. Merge Request
  • Comparing Approaches to Durability in Low Latency Messaging Queues

Comments

Database Partner Resources

X

ABOUT US

  • About DZone
  • Send feedback
  • Careers
  • Sitemap

ADVERTISE

  • Advertise with DZone

CONTRIBUTE ON DZONE

  • Article Submission Guidelines
  • MVB Program
  • Become a Contributor
  • Visit the Writers' Zone

LEGAL

  • Terms of Service
  • Privacy Policy

CONTACT US

  • 600 Park Offices Drive
  • Suite 300
  • Durham, NC 27709
  • support@dzone.com
  • +1 (919) 678-0300

Let's be friends:

DZone.com is powered by 

AnswerHub logo