It is official – since a couple of weeks I have become a co-leader of one of the newest Virtual Chapters of PASS – In-Memory Virtual Chapter.

What is In-Memory? In-Memory describes a set of technologies which are intended to function purely inside of the main computer memory (Cache + RAM) and which are expected to execute extremely fast, since one of the slowest components (and often the main bottle-neck) Disk Drive is not involved.

In practice, right now, it seems that a number of new technologies are commonly described as an In-Memory, even though I do not agree with this marketing term.

Which Microsoft Data Platform components are described as an In-Memory:

– Hekaton (aka In-Memory OLTP) – A true In-Memory OLTP technology, which is right now in the very first version with a lot of limitations but with ENORMOUS potential.

– Columnstore Indexes – Nonclustered & Clustered Columnstore Indexes.

– Buffer Pool Extension – useful technology if you using Hekaton, it extends your Memory space to include the fastest of your disk (SSD or better) to prevent the situation that your Hekaton simply stop receiving modifications.

– PowerPivot – hosting the Vertipac compression engine, this application engine for Excel allows to make blazingly fast advanced calculations, import data from multiple different sources and most of all the possibility to manage & model the logical connection between different tables.

– Tabular Mode from SSAS – oh well, it is a kind of a PowerPivot for SSAS. Going “BigData”, you know

Why In-Memory?

A lot has been said and written in the past number of years, and the truth is – you can have 2-3 TB of RAM right now for a very accessible amount of money, and all those small Virtual Machines with 4GB of RAM are going to fade away in the nearest future (couple of years). The current hardware trends show that we can put the most important tables working directly In-Memory or in some cases the entire application DB can be put inside the RAM.

From the other side the CPU speeds are not increasing (a single-core would melt if the cpu frequency would increase), and so in order to get better performance one should parallelise it’s application code, no matter if it is an OLTP or an OLAP application.

What’s next?

Together with great Ami Levin I am intending to take this Chapter to the place where it truly belongs – as one of the leading chapters in the Virtual space.

We are kicking of on the 26th of November with John Q Martin (Microsoft Premier Field Engineer based in the UK), who will be presenting a very important session, which is called “The Great Buffer Pool Extension Mystery“.

Even though and actually maybe because it is a Buffer Pool Extension session, that is being delivered by someone who is actually using it – I invite EVERYONE to attend it to learn.

In December I am looking forward to make a special event which will include a number of presenters talking about a number of technologies.

I will dedicate part of my time for searching for some reliable sponsors who would be interested in providing some awesome raffle prizes for the 2015.

Stay tuned!