Free Month of Plural Site
I'm interested in seeing your responses. Any relevant response will be accepted and a free month of Plural Site will be given away. You do not have to agree with me to be relevant. I want to hear what you think.
This is a bit long winded for me.
First we will address the minimum requirements. We are comparing core licensing only. You must purchase at least 4 cores. Enterprise Edition costs $6,874 per core. That's a starting price of $27,496. Standard Edition costs $1,793 per core. That's a starting price of $7,172.
Now I know what you're thinking, a $20,324 difference seems pretty open and shut against this. This is where I ask that you hear me out. It gets a bit tricky here. Let's start comparing the benefits of upgrading.
I know there are a lot more difference relating to BI, AS, RS and many other aspects. Let's just get enough out there to prove the point. I just told you that a $20,000 dollar cost was a savings yes? How can we save $20,000 by spending it?
How much do you make a year?
How about your other DBA, or the JR you're about to hire? How many Developers do you have on staff? How many of them are over worked trying to keep your old Standard server running? Look at your database closely. Let's answer some questions; we'll address this question last.
Do you have the maximum ram that your server can support in it?
That may be 192, 384, any other number. Unless your server is older, it should support more than the 64GB of RAM that Standard does. I know I know, Windows Server Standard only supports 32GB of RAM, but that changed in Server 2012. 2TB is the RAM limit now. I know SQL 2014 allows for 128GB of RAM standard, but that still means more room to grow.
Do you have processor cores just going idle most times?
Just because you have 12 cores doesn't mean you need to license 12. You can set SQL to use the limit of what you license it to use. Only license what you need.
Do you have SLA's to meet that have been difficult due to maintenance windows?
Online indexing allows you to rebuild tables just about any time. You do still get a minor lock at the start and end of an online operation, but that's far better than during the whole process.
Do you have issues with archiving those massive tables?
Is their performance falling behind? Partitioning can help you swap parts of the table in and out while being minimally intrusive. You can even address fragmentation per partition instead of hitting that 10 billion row table all at once. In 2014 you can even do that operation online now!
Do you have multiple databases on the same server fighting for resources?
Well now you can split them up logically instead of having that same conversation about splitting them up physically.
Are you fighting for more space or even considering moving to an Enterprise SAN?
Page compression is a beautiful thing. It should pose no problems on archived tables. If CPU is not a bottle neck currently, you can expand the window of what you compress. Heavily used tables may not benefit as much... but here's where partitioning can work with this. Page Compression saves quite a bit of space. This may be just enough of a space saver to allow you to request those SSDs you've been wanting.
Let's add this all together.
Yes there is a cost up front, but now you no longer need to hire a 3rd full time DBA or Developer. If you get that system on SSD's since we're using page compression to keep our sizes small and partitioning to keep our archived data on slow disks, our response time is faster. We were previously running to the edge of our RAM at all times, now we have cached static data from a month ago. These disks are only being accessed to present changes. Our DBAs are getting more sleep now that their fragmentation jobs aren't blocking all night long. Our maintenance windows are getting shorter and our SLAs have more room to breathe.
This view will not fit all organizations. As always, the phrase "It depends" will fit in this scenario as well. Think carefully about all these issues and the time you've burned fixing them. You could be working on that next project to make your company even more money.
I mentioned Developers in this post. I'll explain now why. With compression, partitioning, more caching, faster access to those tables... you can hide a lot of "quick" codding with that much faster power. I'm not saying that we should code poorly because we can. I'm saying that we can code how we need because we can. Once it's up and working, you can then go back and fine tune.
Enterprise Edition has a higher cost. Asking for more SSD's and more RAM has an additional cost. Not having your talented knowledgeable DBAs and Developers quit due to long hours, continually fighting uphill battles and being denied tools or extra personnel will cost you a lot more in the end. Training your next DBA alone may make up this cost. Not losing your customers due to the inability to meet SLA requirements has a large cost monetarily and to your reputation.
Let's make the world happier, one Database shop at a time.
Post a Comment
All opinions welcome! Please comment with any changes thoughts or discussion points.