June 18, 2012

Introduction to MemSQL

I talked with MemSQL shortly before today’s launch. MemSQL technology basics are:

MemSQL’s performance claims include:

MemSQL company basics include:

As you might think, MemSQL’s technical story is focused on performance. Basics there include:

The idea of compiled queries is hardly a new one, having been raised before by (for example) QlikTech, StreamBase, and ParAccel. Highlights of MemSQL’s version include:

Queries only take up a couple of kilobytes each, and in the early going MemSQL hasn’t seen customers with more than a few thousand stored patterns. Hence the MemSQL guys are highly optimistic that storing every query pattern a system has ever compiled will not create any material space crunch, and just have a simple LRU (Least Recently Used) approach to freeing space in that unlikely eventuality.

MemSQL’s persistence story is that you flush write-ahead logs to one set of disks (spinning or solid-state), while sending snapshots (perhaps continually) to other disks. Persistence is designed to be sequential, although I’m not sure whether that remains true in the full-durability case.

Other random technical notes include:

Related links

Comments

3 Responses to “Introduction to MemSQL”

  1. Vlad Rodionov on June 18th, 2012 1:58 pm

    For HFT application 1 ms is an eternity :) . When someone speaks about RDBMS “insert performance” one detail which is almost always missing is the presence or absence of table indexes. But I suppose they do not do any indexing except by some primary key (for skip lists).

  2. Leigh on June 19th, 2012 2:41 am

    When ever new technologies come about in the data space, thousands of people scream “this will be the death of SQL”. Then SQL vendors evolve and things go back to normal. Mark this as the beginning of the end for the NOSQL movement.

  3. Igor on June 24th, 2012 12:08 am

    What’s the avg. record size and # of fields/record
    in the benchmarks mentioned above?
    1.2M inserts/sec is not very impressive if records are 10 bytes – and very impressive if they’re (let’s say) 500 bytes. So what’s the story?

Leave a Reply




Feed: DBMS (database management system), DW (data warehousing), BI (business intelligence), and analytics technology Subscribe to the Monash Research feed via RSS or email:

Login

Search our blogs and white papers

Monash Research blogs

User consulting

Building a short list? Refining your strategic plan? We can help.

Vendor advisory

We tell vendors what's happening -- and, more important, what they should do about it.

Monash Research highlights

Learn about white papers, webcasts, and blog highlights, by RSS or email.