SQL Server OR Postgresql
-
Hi,
here is my problem or, may be, it sounds more like a joke...
We are running SQL Server 2000 (Enterprise and Standard) on Win 2000 & 2003. Bought licensing and everything.
Although as DBA I do not have any serious issues (my major problems are when SQL backup doesn’t find the network path – bad switch, or domain problems – PDC down), my boss thinks “SQL is unstable” and proposed to switch to Postgresql and eliminate SQL Server. The reason – “it’s for free” and it’ll be “more stable”…
I did look at the FAQ and Manuals on the postgresql website and it did not look any better to me. The worst is that there’s not a way to compare – no concrete data…
For example: It says: ”can use concurrent connections”, but how many, on what hardware, etc. – it does not say…
Anybody using postgresql? Somebody knows some articles for comparison?
Any thoughts are greatly appreciated.
Thanks,
MJ
-
I looked at this a couple years ago along with MySQL. It's been a few versions and the PostGres/MySQL versions have advanced quite a bit.
My 2cents:
SQL Server is not unstable. I manage the databases for Peoplesoft internal, a 12,000 person company and I wouldn't hesitate to run any of our enterprise apps on SQL Server. It runs well, is extremely stable and in many cases outperforms our DB2 installs. Oracle is still the leader for performance and scalability, but it's a pain, expensive, and we don't need that scale. SQL Servers big selling point is ease of use and quickly getting things done. Also lots of features, but most people (including me) don't use most of them.
MySQL - less features than SQL Server, but for many people, you might not need the features. Not sold on the recover abilities if you have a serious crash, but SAP is, so it must be pretty good. No stored procedures (yet), but good enough for many things.
PostGres - kind of in the middle. More features than MySQL, less than SQL Server. Very stable, ran well, a little cumbersome to get working, tools less friendly.
All of them are good enough for most things. And I mean the vast majority of things. However, aside from the software cost, NONE of them are free. There is a learning curve, admin time, resource cost to all of them. And this is by far the biggest cost. If your developers take twice as long to build something on Postgres v SQL Server have you saved $$ in licensing? Maybe, that part is something you have to assess, but there still is a cost.
You can get support for any of them, but that has a cost as well. When your db is down at midnight and has to be up by 7am, do you want to post on the Internet and hope for an answer? Probably not, so you'll pay for some type of PostGres support company.
I used to not believe this, but I have learned the licensing cost is a small part of the ongoing costs. What if you become the Postgres expert and then quit? How easy is it to find someone with Postgres experience and have them learn the system? What risk are you taking (soft cost)? If it were me, and I don't know your situation, I'd go with MySQL before Postgres because it's more popular and it would be easier to find support, hire people, etc. However, if I'm a Windows shop, then I'd probably stick with SQL Server. It's integrated nicely with Windows servers and desktops, drivers already installed, etc. And extremely stable. I only reboot servers for patches. And those are usually 90-120 days apart, less these days. I have a 600GB data warehouse on SQL Server (4x4) that outperforms our 200GB DW on DB2/AIX. Most of that is because of the design, but it does show that SQL can do as well as DB2. Maybe it's a stretch, but I think it can easily perform as well as Postgres.
-
Steve's got a great answer, but I'd like to throw in my 1cent -- after Steve's reply, that's all mine is worth
.The point about licensing being a small cost in the overall cost of operations is a critical one. In responding to your boss, I would explore the estimated costs of getting up to speed on Postgre or MySQL (and I agree with Steve, go with MySQL if absolutely necessary; more avenues for support and existing skillsets). That would be training time (probably self-taught), time to evaluate the MySQL feature set, time to convert the system from SQL Server to MySQL, establishing new backup plans, etc. While most of this would be done in house, my gut tells me that you'd need to hire a consultant to at least aid in direction. The cost of that consultant only would probably offset licensing costs alone.
Also, if you already have SQL Server, you've already paid for the licenses, so converting is throwing money down the drain! If it's a scenario where new CALs are needed, at a certain CAL count (I think about 25 for standard edition), processor licensing is more cost-effective. And you don't have to worry about adding new CALs and such.
As far as "unstable", in what way does he mean? Are there actual incidents that he's referring to? Or is it just something he's read? That might be a tricky conversation to have with your boss, but that would be the first thing I would ask as a consultant.
And one last thing -- the backup issue you described. It's my personal preference NOT to backup to network devices. I prefer to backup to disk, then to tape or copy the backup file to a network location. My feeling is that the backup process takes longer to execute than a file copy, and the longer a backup runs, the higher the likelihood of problems. Just my take on it.
Good luck with your situation!
-
For the backups, which I didn't address. The backup process is finicky. Any network delays will cause failures and they need to be rerun. Backup to local disk and then copy.
-
Hi,
Backups are integrated with both MySQL and PostgreSQL to TSM nowadays.
Take a look at:
http://www.repostor.com if you are interested.
Data Protector for PostgreSQL using IBM Tivoli Storage Manager
and
Data Protector for MySQL using IBM Tivoli Storage Manager
Regards Tomas
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply