Insights

Why haven't we seen an SAP HANA vs Sybase IQ showdown?

13 Jun 2011 Business Intelligence (BI), HANA, In-Memory, Enterprise Architecture, Consumer Business

You may be surprised to see SAP HANA and Sybase IQ mentioned in the same sentence.  I's certainly not something I've come across very often in SAP material or conferences, and while I can understand the reasons behind this I think it's quite unfortunate. Both of these tools fundamentally solve the same problem - they enable organisations to query massive amounts of data at impressive speed. HANA stores the data mostly in memory and IQ stores the data mostly on disc - other than that the technologies are not that different.  It seems obvious based on this that between the two HANA will be the winner on performance. No arguments there.

However, other than this basic acceptance, we have very little (if any) facts to go on that will allow us to compare the two solutions.  Again, based on the physical architecture it can reasonably be assumed that HANA will be more expensive than IQ - and no doubt this will be true for software costs as well. So HANA is better and IQ is cheaper.  Is that it? Hardly enough information to make an investment decision on.

Big data continuum

SAP does not view these technologies as overlapping, but rather as having distinct places along a "big data continuum". This is the explanation I got from George Mathew at the recent BI4.0 launch event in Johannesburg. To clarify this he mentioned, among other things,  that Sybase IQ can handle petadata-scale databases, which HANA cannot. So SAP's position is that there is a place for both of these solutions and they don't compete. This goes some way to explaining why SAP doesn't see the need to present any comparative statistics. 

However, I think this is an over-simplified view of the world and that it is pretty obvious to most people that have some understanding of both solutions that there is at least some overlap. The "big data continuum" concept is also extremely vague and I haven't seen any detail of this that could help customers make investment decisions (the best summary I've seen is in a blog by Helena Schwenk). This perspective also assumes that IQ (or HANA for that matter) only comes into the picture with "big data".  What about "medium data"? It doesn't take that much data to start experiencing performance problems - a 100m row dataset will do.  You may not consider implementing HANA in this scenario, but should you consider IQ?

Lost opportunity

SAP has a very clear marketing objective with its HANA communication and including comparative Sybase IQ statistics may not necessarily have the desired effect. It is very clearly targeting only "big data" scenarios at the moment, and in a way I don't blame them as this makes for the most compelling stories.  But it does feel like this caters for an elite group of customers only (i.e. organisations that could afford HANA).  The problem is that medium sized organisations with slightly shallower pockets also have problems with performance - problems that could potentially be addressed by IQ.

HANA v IQ - why should we know?

For me it's all about understanding the options and helping customers make the right decision for them.  Here are a couple of points to consider:

• SAP HANA is a brand new product on v1.0 with no production reference sites(yet) - IQ is a mature product (v15.3) with hundreds of world-wide reference sites
• SAP HANA does not yet offer Disaster Recovery or High Availability features, which IQ does
• There is not much SAP HANA skill in the market so it will be hard to come by and pricy - equivalent IQ skills will be easier to get hold of and much cheaper
• SAP HANA is likely to require more expensive hardware and have higher licence costs

It's clear that IQ has a lot going for it. However, the point of this is not show the superiority of IQ, but rather to bring some balance and reality to the HANA discussion.  No doubt HANA has tremendous promise and will be a fantastic enabler in many scenarios. But it has some way to go before it is a mature product and in the mean time options like IQ deserve some consideration.  Of course the missing element to the list above is the answer to the following question:

"If I pour the same large data set into HANA and IQ, how does the query performance compare between the two solutions? I.e. if HANA comes back in 0.4s, does IQ take 2s? 10s? Or 1min?"

Nobody seems to know. Or if they know they are certainly not telling. This is a situation that needs to be remedied and all it will take is an IQ installation (done) a HANA appliance (hopefully arriving soon) and some test data.  Watch this space for a blog on the outcome!