CTP 2 of SQL Server 2014 was released last week and I wonder if there is anything worth upgrading for if you are coming from SQL Server 2012.
There are two things that look interesting to me; these are updatable ColumnStore indexes and Extending Memory to SSDs
In-Memory Updateable ColumnStore:
Provides higher compression, richer query support and updateability of the existing ColumnStore for data warehousing workloads giving you even faster load speed, query performance, concurrency, and even lower price per terabyte.
Extending Memory to SSDs:
Seamlessly and transparently integrates solid-state storage into SQL Server by using SSDs as an extension to the database buffer pool, allowing more in-memory processing and reducing disk IO.
Hekaton: OPTIMIZE FOR CRAP?
During the PASS Hekaton demo a horrible designed database was shown to greatly benefit from using Hekaton. What I and other people wondered was if Hekaton would have helped at all if the database was designed in best practices way. Here is what Brent Ozar wrote during his liveblog event at 8:55 AM: “Oh, this demo fills me with rage and sadness. Everything about the tables and the queries is worst practices. This has absolutely no business in a production database and an online store. If your site and your code is this bad, and you refuse to use a cache, then sure, Hekaton is for you.”
I agree with him, of course you will have people say that hardware is so cheap these days that there is no need to optimize (meaning design with best practices) anything, throw more RAM, CPUs, SSDs at it and it will perform as you envisioned it.
Hekaton could be useful for things that change very frequent, for example if you are displaying real time tick data from stock exchanges, you don’t need to store this per se, if your server goes down, you can just pick up the latest tick from the exchanges again. Then again Streambase or Streaminsight might be better for that stuff as well.
Another thing to be aware of is that Hekaton has a lot of limitations, you can’t use certain data types, you can’t use certain operators etc etc. It is not as simple as they make it out to be…schema changes will be needed, we all know how difficult those are
SQL Server book programming authors get an almost free pass.
It was disclosed that CTP is feature complete and that this is the final CTP before RTM. There is very few new T-SQL functionality that will be shipped in SQL Server 2014, so what are the book authors to do? Sure they can do a chapter or two about Hekaton and most of the other stuff will not change. But anyone who already has the 2012 edition of this book will just download the SQL_Server_2014_In-Memory_OLTP_TDM_White_Paper.pdf to get up to speed with Hekaton
So what is you opinion, will you upgrade from 2012 to 2014?