If you are looking for a server appliance that is pretuned to load an Oracle database on, then Ellison & Co have a new Database Appliance X3-2 they want to introduce you to. Not everyone needs an Exadata database cluster that can scale to petabytes of capacity, and that's why Oracle engineered the Database Appliance for small …
When you need high performance....
The data professionals choose Linux. This box is another example of this. Furthermore, it's another example of how the TCO of linux is better.
We better get used to Exa being the new "Mega" and "Giga". It seems like "Peta" doesn't sound quite as cool. It sounds like a bunch of ugly, anorexic models that eschew fur.
Re: When you need high performance....
Why would "Peta" sound better?
1) How many actually like PETA?
2) Peta and Pedo sound quite similar. So if some makes a filer, would someone buy a PetaFiler?
Re: When you need high performance....
You can also run Solaris on Exadata if you wanted.
I hear the MTBF of those Sun nodes is about 2.5 years which is the #2 reason we have steered cleard of exa-xxx. The number one reason is they are exa-expensive and the single vendor lock-in is with the worst IT vendor out there as far as price and terms.
cheers to my ex-sun co worlkers
"...I hear the MTBF of those Sun nodes is about 2.5 years..."
Do you have any links or proofs to back this up? Considering your long history of FUD and lies, this is also most probably a lie...
> Two of those eight bays are used for a pair of 200GB SSDs (again for the database redo logs)
Who uses SSD for redo? Redo logs are sequentially written, so this is waste of cash. There is no random IOPS! All other storage vendors recommend using 15k HDDs for redo logs. Much chaper, same performance.
Heh, disagreeing with the advice of the people who wrote the software, I like your style sir.
... and randomly read for MVCC.
I can (hardly, but still) understand reasons for buying exadata. But I am just not able to understand what kind of moron you must be to even consider buying this appliance. This is nothing but commodity hardware with premium price. You can get exactly the same setup (and better MTBF, not even talking about support quality) for half this money by buying for example hp proliants. Or IBM stuff. Or Dell. Unbelievable that there are any customers for this box.
They do use a fair amount of commodity hardware in Exadata, but since they control the database, they're able to move some of the database "functionality" down to the storage level. The performance offered by this is one of the reasons for the price premium.
As to who would buy an Exadata? The primary folks are those with large or multiple oracle databases that need performance. As well as those that want to lower storage costs. It's also highly likely you'll be able to lower the database licenses by flipping to Exadata. By being an appliance you also look at saving admin costs due to not needing as many people taking care of the servers, network, storage, or databases. So when you factor in the TCO, that price premium is not as high as it may seem.
As far as other vendors. We put forth a major re-compete across 6 different vendors. Oracle was able to clobber the other vendors with Exadata in not only performance but also cost. That being said. Oracle is not a terribly customer friendly vendor, but if you have a use case for Exadata, it's a hard sell to not put an order in for a few racks after seeing what it can bring to the table...
As I said (and you didn't seem to notice) - I can understand the reasons why some people buy exadata. it really contains couple of proprietary solutions on storage level, true. They are sometimes useful, more often not, but they do exist. But they are in exadata, NOT in this appliance. This appliance is nothing but commodity hardware and plain oracle linux over it. And cost at least double price.
"the database "functionality" down to the storage level."
Look at how simple the queries need to be to take advantage of Smart Scan. It is a really small portion of your queries that are even applicable. All of the other functionality has similar caveats, HCC has some benefits for bulk loads, but is not applicable to row level inserts and the like (i.e. it is only for OLAP).
"It's also highly likely you'll be able to lower the database licenses by flipping to Exadata."
I have heard this, but can't believe it. Think about that from Oracle's perspective. Are they going to trade extremely high margin software dollars for relatively low margin hardware dollars with no 22% maintenance contracts? They are not doing this to lose money. You are also forced to use EE RAC licenses, the most expense Oracle database license available, regardless of whether or not there is a benefit for a particular workload.
"By being an appliance you also look at saving admin costs due to not needing as many people taking care of the servers, network, storage, or databases."
Yeah, but it is not a black box appliance. You still need to admin all of the Linux OSs, all of the DBs, all of the specialty software. Oracle tells DBAs that they should go to an 8 day class to learn how to DBA an Exadata and there will often be a lot of query rewrites or tuning to align with their unique implementation of RAC.
There are many reasons to buy this.
1) You have idiots for your storage admin and system admin. That no matter how many times a dba tells them exactly how to configure something, they screw it up. This box eliminates the morons. It gives the dba everything under his control.
2) Licensing consolidation. I've seen too many cases of database sprawl at SMB's. For a manager, he knows if it's on this appliance, it's licensed. And if he really needs to he can expand the cores and licensing without have to do another costly (in dba time) hardware migration.
3) Ease of deployment. Having demoed one, it literally only takes a few hours after unboxing to having a database running. There is no way you could setup a two node oracle cluster this fast on a system you build, unless you build 100 of them and get it down to a science.
4) Patching. Firmware and oracle patches are bundled together, supposedly tested and certified to work. Patching is very simplified as well.
5) Disaster recovery. Say you want to setup a D.R. site but don't cannot afford, or convince management, to duplicate your oracle licenses. You can log ship to your mission critical databases on an offsite Database Appliance with limited cores enabled. If you ever DO actually need it in a disaster scenario, you can up your cores and licenses rather fast and you're back in business.
6) Bundled third party software with the database appliance. Vendors have the ability to take a database appliance, put their application on it, and bundle it with the oracle database. Then they can sell an all in one solution with built in high availability and growth potential.
I could go on and on. The only people that fail to see the potential with this device are, frankly, snobs. This device is definitely one of their better offerings. It has the most flexibility and utility of any Oracle product I've seen in a long time.