FusionIO vs. Hana?

[Former Member]

Hi there,

i am not that HANA Specialist. Our BW System is running at the Moment on FusionIO. So my Supervisor comes to me and asked if we should not go to SAP HANA.

So when we want to use SAP HANA do we have to migrate the whole BW System to HANA? Or is it so that only Infocubes Methods and so on are only stored in the HANA System?

Can someone here explain the benefit FusionIO vs Hana?

Kind Regards

[Former Member]

Hi Bjoern,

I work at Fusion-io and wanted to let you know that Fusion-io actually is used to accelerate many HANA solutions today, so you can actually run HANA on ioMemory as was explained above. For example, you can see news from Fujitsu on their integration here, http://www.webwire.com/ViewPressRel.asp?aId=136044#.UeBS8utX-Fk. Many of our other systems vendor partners feature Fusion ioDrives in their HANA appliances. If we can help with more info, please feel free to send me a note at socialmedia@fusionio.com.

Thanks,

David

[lbreddemann]

Hi David,

David Malmborg wrote:
Hi Bjoern, 
I work at Fusion-io ...

thanks for that!

It's not that often that people make it clear upfront where they are coming from.

Appreciated.

- Lars

[lbreddemann]

... and just comment this:

David Malmborg wrote:
Hi Bjoern, 
I work at Fusion-io and wanted to let you know that Fusion-io actually is used to accelerate many HANA solutions today, so you can actually run HANA on ioMemory as was explained above.

What Fusion-IO can accelerate are disk IO accesses.

That includes things like reading data from disk to memory, writing logging information to disk, writing out savepoint information to disk.

It doesn't speed up the main part of data processing in SAP HANA as this happens all in memory.

It doesn't improve the way SAP HANA processes data in parallel and it doesn't allow for data model changes as SAP HANA does (flat infocubes, less secondary indexes, no aggregates... ).

So to say Fusion-IO accelerates HANA solutions is a bit too blurry to not be marketing talk

It speeds up the part of database work (I/O) that doesn't play such a big role anymore with SAP HANA.

- Lars

[Former Member]

Hey Lars,

Totally agree with you - Fusion-IO improves one specific piece of the HANA appliance. It does this for all single node appliances and IBM scale-out. All other vendor scale-out use SAN solutions for all data.

The redo logs are stored on Fusion-IO. So, Fusion-IO specifically improves the performance of loading data into SAP HANA whilst retaining the ACID properties of the database.

Since you can write at around 1GB/sec to a Fusion-IO card, HANA can load data into memory at up to 1GB/sec. Note that this also means that there is a theoretical improvement in replaying redo logs into memory on startup, but I've not seen that to be I/O bound.

Like you say, Fusion-IO does NOT (in any HANA appliance):

- Improve query performance

- Improve startup time (apart form redo logs)

- Store any data apart from logs

In many other database solutions, you would store your data on Fusion-IO so you improve the performance of cache misses, but this is not how HANA uses Fusion-IO.

John

[Former Member]

Thanks John and Lars for the further clarification.

David

[Former Member]

Hello Bjoern,

Just to let you know Fusion-io also has flash arrays that are available. There are in fact two different kinds, one is the ioN which is an all flash array and then there is NexGen with is a combination of Flash and spinning disks.

Fusion-io also has a software package called ioTurbine, that works on both bare metal and virtual environments and works with a fusion-io card that is located in the server.  The software and the fusion-io card work together to cache all reads from the server with in the server without going to primary storage, and gives you the ability to greatly increase IOPS in read intensive workloads.

All this information can be found on the Fusion-io website.

-Ben

[Former Member]

Hi Bjoern,

FusionIO are typically just adaptors which are running in your server for caching or for log files with very high IOPS. Whilst they have their particular use case I would wonder if this will not become very fast very expensive ?

I have seen now several vendors who are positioning larger Flash Appliances (Violin Memory, IBM etc) in order to connect to a SAN Volume Controller. The idea there would be that you have a mirror of your "normal spinning" disk array just for read access on Flash. This will give you some good acceleration if all you look for is speed. This would be similar to bolting on a Turbo on your current car engine.

Clearly HANA offers you many more advantages and it is worthwhile to take a good look into it.

[chanda]

Hi Carsten Nitschke,

I was just trying to know how violin works and what way it is different from HANA.

Violin promises about 3-5 times faster performance on SAP applications . HANA claims much much faster.

can you please give me some information in comparing HANA & Violin, as our management is looking at both options seriously.

Thanks & Regards,

Chanda

[henrique_pinto]

As already mentioned, you're comparing two different things.

You cannot compare HW x SW.

What Database is your BW system on today? Oracle? DB2?

That's what you gotta compare to HANA.

Also, most of the HANA certified HW uses FusionIO for the log files (read/write), while using SSD for the data files (read).

Best,

Henrique.

[Former Member]

Hi Bjoern,

Just to add to Vivek comprehensive reply you will need to ensure you run SAP HANA on certified hardware. You can get a list of the certified vendors and servers on the PAM.

SAP HANA PAM

[vivekbhoj]

Hi Bjoern,

According to me :

So when we want to use SAP HANA do we have to migrate the whole BW System to HANA? Yes, you have to migrate whole BW system to HANA :

Or is it so that only Infocubes Methods and so on are only stored in the HANA System? No, Everything has to be migrated

You replace your existing Database with HANA.

After migration also you have to convert your DSOs to In Memory DSO and your Infocubes to INmemory Infocubes

Well if you are going to use HANA you are getting In Memory DSO and In Memory Infocube which have a slight structure change as "

You don't have Aggregates, Indexes, Dimensions( except for Package Dimension) and then there is only one fact table in In Memory Infocube

To learn more about IN Memory Infocube :

http://scn.sap.com/community/hana-in-memory/blog/2012/01/11/bw-on-hana-working-of-in-memory-info-cub...

In IMDSO we don't have a physical Change Log table Instead we have a Calculation View and our Active Data Table acts as a "Temporal Table" .

Learn more about IMDSO :

http://scn.sap.com/community/hana-in-memory/blog/2012/01/05/bw-on-hana-working-of-in-memory-dso

To learn more visit cookbook BW on HANA :

https://cookbook.experiencesaphana.com/bw/

Some more blogs :

http://scn.sap.com/docs/DOC-41509

http://scn.sap.com/docs/DOC-33254

http://scn.sap.com/community/bi-platform/businessobjects-bi-for-sap/blog/2013/05/28/how-does-all-thi...

http://scn.sap.com/people/thomas.zurek/blog/2011/06/23/hana-and-bw-730--part-1

http://scn.sap.com/people/thomas.zurek/blog/2011/06/24/hana-and-bw-730--part-2

As per my knowledge :

HANA is a In memory database - whole data is stored in RAM.

Fusion-IO is not a database - it is a in memory device which can be installed and configured in the database to use it for temp files, log files, and indexes to triple performance or we can also move whole database .

But Fusion-io needs a database like Oracle or MySQL to work on.

So HANA is a In Memory Database while Fusio-IO is used to accelerate database performance

There are many benefits to HANA as you get IMDSO, IN memory Infocube.

To learn more :http://www.fusionio.com/solutions/database/

Fusion-io  can be used with SAP HANA also :

http://www.fusionio.com/blog/fujitsu-leverages-fusion-io-for-sap-hana-solution/