@BenBrocka maybe a large scale garbage collection operation to free up some space. Doesn't the TRIM protocol advise the disk that certain blocks are no longer in use, or am I confusing it with something else.
Real-time (i.e. interruptable) garbage collection on RTOS platforms is a long way from being a mature technology; I guess it's possible that the disk could have run into a similar class of problem.
However, this is really no more than idle speculation.
@ConcernedOfTunbridgeWells I wish. SQL server 2005 with PHP on windows, using a PHP framework that only recenlty got proper support for sql server. Getting SS to work for our site has been a nightmare
@ConcernedOfTunbridgeWells Generally Rackspace are a very slick organisation, would recommend them. This case is a new environment build which includes a SQL cluster, so it may be that the right eyes aren't on the case just yet.
Actually, a DL360 isn't much smaller than a folded deckchair. Maybe you could do something like that bit on pro wrestling matches where they scone each other with deck chairs, only use a 1U box instead.
There is a significant risk that people will get the ear of someone in your organisation who doesn't know enough to see through the bullshit. That's how a certain vendor of network attached storage equipment managed to sell Oracle over NFS as a data warehouse platform to my current gig.
Unfortunately, management by reading magazines (MBRM) is pretty endemic in uk corporate culture. This is the 5th outfit I've worked for on the trot that's had this type of problem with inappropriately specified MI platforms.
The only people who seriously advocate this sort of shit for data warehouse systems are high-end storage vendors. Unfortunatley it's next to impossible to get hold of modern SAN kit and do a quantitative benchmark to debunk the bullshit.
@JNK Ebay is your friend. You can buy older server kit - including disk arrays - for peanuts, and do your own experimentation. SANs are a bit of a minefield, though. Often the licenses for the admin software are not transferrarble.
@gbn Actually, the place I'm staying in has central heating. However, the disk arrays used to do a good job of heating my office during winter at one point.
@MarkStoreySmith Well, at one point I had the option of buying a DMX2000 for about 5,000 EUR. Unfortunatley it was two rack cabinets with 192 disks, and I was living in a three-bedroom terrace.
I have one table where I want to check if record exists leave it alone, if not insert new row and update previous row. I am wondering if I can use merge here like below ?
CREATE TABLE a
(keycol INT PRIMARY KEY,
col1 INT NOT NULL,
col2 INT NOT NULL,
col3 INT NOT NULL);
INSERT INTO a VA...
On the plus side, DMX's actually come with a built-in management console, so they don't put you in the business of trying to scrounge a dodgy copy of Navisphere.
@MarkStoreySmith After running 3-phase power into the house :)
@ConcernedOfTunbridgeWells For anything "non-core" to my skillset, that tends to be my approach. I don't need to be expert but I need to be able to spot someone pretending to be one.
Most of my bullshit-wrangling has been through infrastructure architects who have already drunk the vendor cool-aid. Unfortunately in the contracting racket, most of the design decisions have already been cocked up long before I get anywhere near the project.
Confidence in Rackspace restored... "I am glad to inform you that our Segment Support Infrastructure team is already aware of this and they have confirmed the "DATA" folder and any .MDF / .NDF / .LDF files are excluded by default for being scanned."
I frequently come across the situation where I want to pass an array of Ids as a parameter to a stored procedure.
For example with GetCustomerAccounts stored procedure, the client chooses say 20 customers from a list. I want to then pass these 20 customer Ids to the procedure for it to return th...
I've got a process that grabs a bunch of records (1000's) and operates on them, and when I'm done, I need to mark a large number of them as processed. I can indicate this with a big list of IDs. I'm trying to avoid the "updates in a loop" pattern, so I'd like to find a more efficient way to sen...
@JackDouglas judging from just the title they look like related but independent questions; however the bodies and answers show these questions are basically the same. any answer on one is a valid answer on the other. don't you agree?
but I don't think it's a big deal; they are already linked to each other so if you want to leave them open then that should be fine
@gbn Yes, Sadly, I think databases are a big enough topic that they really need to be a primary speciality to do well. However, there's nothing stopping a database specialist from knowing a bunch of other stuff as well.
Interesting. I don't want to sound critical, but SO database isn't that big and hasn't as much traffic as some that are mentioned here,. I always thought the main issue they had was the web and middle tiers, caching, DNS, SSL et. Not the DB per se.
I think they do OK to run it on fairly modest hardware though. The server has a lot or RAM for caching, but otherwise they've not had to bother with a big hardware footprint.
Inverted indexes aren't super efficient though - you still have to stuff your results into a priority queue to get the highest cosine ranking out quickly.
@MarkStoreySmith Reminds me of my third year software engineering project - hacking a full-text search system to change its definition of 'word' to encompass search terms we wanted to support on javadoc collections.
@NickChammas only after a period of time. Until then it remains alive with a pointer, much as a closed Q does.
@JNK just mentioning that it's not totally foreign or impossible for eventual consistency to work on such a large site. Would be awkward with SE Q and A, but not impossible
@NickChammas The old box becomes a performance testing box then. Not wasted money, really. Example, spank in more questions and comment to predict growth and load patterns
the long and short of the current frustrations is that she knew she was leaving town with her mother for two weeks and didn't do any laundry till two days ago, and yesterday when we should've left the house at 5pm for her to go to her mothers in preparation for leaving at 1am, we didn't leave till 9:15
I think you would have major issues with BASE on SE....not knowing if an answer was already accepted. Not knowing if something has been flagged. Not seeing if others have VTC a question. Multiple bounties on same Q. Answering a closed question, or deleted question, without knowing it was closed/deleted.
Stack Exchange co-founder Jeff Atwood announced that he is leaving the company to spend more time with his family, including his twin daughters Maisie Jane (5lb2oz) and June Adeline (5lb 7oz) who celebrated their 0th birthday (and joined Twitter) last Friday, to the great joy of their parents.
It has been a great honor for all of us to have worked with Jeff over the last four years as Stack Exchange grew from absolutely nothing to a world-changing resource with over 30 million monthly visitors.
When I first met Jeff, I told him that when Stack Overflow was built, it would become a standard part …
@NickChammas Which is curious for me. Anyone who can scale out can afford racks/power/cooling and custom solutions. But most folk can't or won't afford that. So scale up, easier for most sites
Some scenarios for M:M relationships in a data warehouse model
Most OLAP servers and ROLAP systems have a means to deal with M:M data structures now, but there are some caveats about this that you will need to pay attention to. If you do implement M:M relationships you will need to keep an eye ...
Really I'm thinking about a big HP server as a RDBMS server box. Big data types can go with NoSQL solutions, and Big data types with money can use Netezza or Teradata.
A DL980 could handle a pretty substantial transaction load on an ordinary transaction processing system workload.
Maybe I should put that on a business card.
Concerned of Tunbridge Wells Bringing intelligence to the business since 2001. cotw@fatbastard.com
@MarkStoreySmith I'm not seeing why the intermediary makes anything safer than setting up proper perms + those data accessing procs on the web database
That's what I've been doing on our set up, moving all actual views and procs onto our web database. The old set up had it's hands in a vendor DB and a company DB directly which I didn't care for
I'm impressed by SSMS's ability to suggest (and use) indexes that cause a query to perform worse
It's 1/4 for suggesting helpful indexes for this app...which is annoying as they take 20 minutes to build. They seem logical enough but perform like crap
I'm looking for a good, well-written and explained guide on reading and comparing execution plans in SQL Server. This is for more my own edification as well as for sharing with others.
If you want to add your own summary here that would be welcome, though I expect the topic is large enough that...
You guys think it's bad form for me to berate askers who just post a list of requirements without any sample code/data or any display of an attempt to resolve on their own?
I've got a CSV that has multiple delimiters and the following format:
groupname;user1,user2;user3
groupname;user1,user2;user3,users4
How can i add the users to the AD groups. All the group names in the CSV ends with an ";" seperator and the users users use the "," seperator.
@BenBrocka Found the one I was after, which includes the DBCC AUTOPILOT command so the optimiser evaluates hypotheticals blogs.solidq.com/fabianosqlserver/…
Hadn;t heard of the autopilot either. I'll have to try this stuff out. This app definitely needs indexes. It's still the slowest thing on the site (fortunately only one person will use it)
@MarkStoreySmith Being able to test indexes without building them is pretty useful...
True...but damn, blindly trusting SSMS's recommendations doesn't seem like a good idea, at least for big tables it seems to give very varied results
Testing a sample query against it's varied suggestions I've gotten query times from 1m20s (base time), 2m+, 50s and as low as 2 seconds
oddly enough the covering index did very poorly and the current non-covering index does way better, even though the base table is huge and the subset is relatively small (200k vs 3m)
shouldn't covering indexes be faster when they're possible?
So, I'm supposed to figure out how to deploy SSRS reports and I'm honestly confused. The idea would be to have the reports viewable through sharepoint mainly to make permissions easier to manage. I don't understand if I need to setup something on SQL server and sharepoint would connect to that, or if I'd just install reporting services for sharepoint and deploy straight to there. We have SQL Server 2008 and Sharepoint 2010.
@MarkStoreySmith Yeah, that's the goal. I apologize if this sounds stupid. I'm a bit of an accidental dba, and sharepoint is definitely not my specialty. So this specific project has been a bit overwhelming. I'm guessing that's what I need to be able to deploy directly to sharepoint. I'll read the link you posted, thanks.
@EugeneM If you've got any specific questions, there is a stack exchange sharepoint site that looks to be well covered with experts these days sharepoint.stackexchange.com
At the moment, I just needed some direction, so that link should be good. I've been struggling to find a low level article that just says "Here are the options for serving ssrs reports." The stuff I've found is more "Here is the standard report server setup. Here is the setup for a failover cluster for report serving."
Also, airlines are sneaky fuckers. They say "no service fee for flight changes", then go to the change booking page and they tag a $50 change fee on. Re-check the conditions and it says in small print at the very bottom "Change fees may apply to some booking classes"
I have one table where I want to check if record exists leave it alone, if not insert new row and update previous row. I am wondering if I can use merge here like below ?
CREATE TABLE a
(keycol INT PRIMARY KEY,
col1 INT NOT NULL,
col2 INT NOT NULL,
col3 INT NOT NULL);
INSERT INTO a VA...
Some scenarios for M:M relationships in a data warehouse model
Most OLAP servers and ROLAP systems have a means to deal with M:M data structures now, but there are some caveats about this that you will need to pay attention to. If you do implement M:M relationships you will need to keep an eye ...
I've got a CSV that has multiple delimiters and the following format:
groupname;user1,user2;user3
groupname;user1,user2;user3,users4
How can i add the users to the AD groups. All the group names in the CSV ends with an ";" seperator and the users users use the "," seperator.
