Hands-On: The Fastest DBMS In The World

The wild world of IT consultancy is full of swings and roundabouts but maybe the best pro from a technology point of view is getting to get to work with a broad range of interesting technologies on the clients behalf. Recently I was on a client site when they got a team from Actian to do a POC on premises of Vectorwise – an very high performance DBMS that I had never encountered in the wild. The best was after they had set up the POC, we (people on the client site) were going to be able to play with it in order to compare it against other alternatives.

First it is important to understand what Vectorwise does well – it is not a database cure-all that does everything excellently. It does analytical style querying well, very, very well. The industry standard point of reference for database performance is the mundanely titled Transaction Processing Perforormance Council (TPC). They have a handful of different comparisons of different database workloads. The most relevent within the data warehouse/BI space is the TPC-H benchmark:

This benchmark illustrates decision support systems that examine large volumes of data, execute queries with a high degree of complexity, and give answers to critical business questions.

This basically gives us our data warehouse style world championship leaderboard of performance and cost. It looks like this:

TPC-H RESULT (2)

Got your attention yet? Thought so.

Cutting to the chase of our experiments on the client site, Vectorwise was running the heaviest queries the finance team relied on up to 100 x faster than the SQL Server being used already. This wasn’t entirely fair as Vectorwise sat on SSDs and had a bit more memory but those aren’t going to make SQL Server 100 x faster. Less intensive queries ran around 10 x faster than SQL Server, sometimes a little less. The Vectorwise consultants told us a little about the way the CPU cache is used exclusively for processing, which is what gives it it’s large performance advantage. Impressive stuff.

Related: Gartner Data Warehouse vendor magic quandrants

We found problems though, firstly run away queries, which the Vectorwise guys came in and promptly fixed through configuration changes.

The largest problem for the client was potentially user adoption as the variety of SQL used was a subset of ANSI SQL that lacks procedures, functions and data-types – the client site’s analysts (the people that would be writing the SQL) used T-SQL and Informix. Re-training a large team rarely seems like an inviting task.

The final problem was an operational one – the particular server we had set up for us was on a Linux server which would be great if there were any other Linux servers in the company or people skilled in Linux administration. There was a windows flavour available which would have eradicated this problem but was not forthcoming for mysterious management reasons.

It seems like a product that on the one hand provides jaw dropping query performance but only the other gives a slightly buggy, feature poor experience. If it can iron out these headaches before it’s competitors can provide the same performance then Vectorwise will be a lot more prominent in the DW space.

Advertisements

About Alister McPherson
Senior consultant at uxcps.com.au, focusing on enterprise business intelligence platforms.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: