By Dr. Berg
Sapphire/ASUG annual conference 2011 in Orlando is almost over. So what is the key buzz? Hana, Hana and Hana... It has been all about in-memory processing.
Let us look at a few details that SAP has provided in the sessions and demos this week:
The New Version Name
I know this is confusing, but SAP has again changed their naming strategy and HANA version 1.5 no longer exist. It is now called Hana 1.0 Service Pack 3 (SP-3). And yes, this will allow you index from SAP BW, unlike the lower level that only supports indexing from ECC.
HANA Pricing
HANA has four cost components. First you need hardware. There are six certified vendors and Cisco just released a cool new hardware appliance at ASUG based on their new C-series servers, but others also have offerings that are cool. I tried to get some numbers from vendors such as EMC, but it was hard to pin them down to even ranges or any other form of estimates. But our friends at SAP (Torsten Hoffman) was much easier to pin down in his public session on high performance analytics. Mr. Hoffman said that customers should plan to spend about $120-$140K for each TB of memory in HANA.
Second, you will need to pay licensing costs from SAP. This largely depends on your organization's relationship with SAP and existing contracts, so talk to your SAP account executive. Also, SAP currently license HANA based on increments of 8GB memory blocks, the number of users and the system purpose. However, a good cost planning number is between $1 and $2 million per TB of memory. Sounds like a lot, but most HANA systems will not initially have 1Tb M emory.
Third, you need an implementation partner to help with the architecture, disaster recovers, sizing etc. And finally, you need to buy a service contract, typically for 3 years to support the HANA environment.
Sizing the HANA Environment
SAP has reported compression of at least 1:5 and a maximum compression of 1:25. That means that if you have a 10 Tb system, you would need at least 2Tb, memory (1 to 5: compression). However, to make sure that there is enough room for temp files, SAP recommends that customers size their environment at 50% of raw data size.
For our example, this means that by only looking at raw data in our 10TB warehouse, we may have only 3Gb 'unique' data. It is this that is our base sizing parameter. The rest of the BW data may be PSA data, replicated DSO or InfoCube data, aggregates and log files that we will not place in-memory. So following this logic, and SAP's recommended sizing strategy, you would need 1500 GB memory (50% of 3Tb).
[Added 6/9/2011: to clearify some questions I received on this sizing approach, let us take a good look at the math example:
3Tb raw data and 1:5 compression = 600 Gb (or 20%)
Double this for the 'free space' = 1,200 Gb (or 40%)
Add 25% room for growth/sizing err = 1,500 Gb (this is 50% of 3Tb)
PS! This has nothing to do with BWA and free RAM. It is a rule-of-thumb to account for compression. Naturally a more in-depth sizing effort shold be undertaken when ready to buy, but this g ives you and idea of what is needed].
Column Vs. Row based Indexes
HANA can index data in both rows and columns, and that is important. Most traditional databases can only index data by rows. That means that if you want the total amount for a region, the database will read the whole row, reject those not in the region, and then present the data. However, if the index is by column, the system only read the data in the column (and filter by org). This is much faster for BI. So, a quick rule-of-thumb: row based indexing for ECC, and column based indexes for BI or operational reporting (but not always, as it depends a bit how you are looking at the data)..
Summary
I know, we have talked about HANA for almost two years, with visions and dreams ever since SAP bought Sybase. But now, here in Orlando Florida, we have seen it work in practice and with real customer data. The conclusion is that HANA has the potential to make a change how we view BI and ECC data.
So, for now HANA for transactional real-time operational reporting, and BW/BWA for decision support. But as SAP adds business content into HANA over the next few years, BWA may become obsolete. It all depends on how fast SAP can get companies to adopt HANA and how much they invest in business content and front-end tools. Today, front-end tools are somewhat limited to Excel, BO Explorer, the statistical programming language “R”, and MDX compliant tools.
It is going to be a fun year... D. Berg