Thread: PostgreSQL giving jitters to Skypak
PostgreSQL giving jitters to Skypak
From
"V i s h a l Kashyap @ [Sai Hertz And Control Systems]"
Date:
Dear all, This story is a blow for PostgreSQL http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&source=cxtnl&userid=aspire420@hotpop.com thoughts ? Best Regards, Vishal Kashyap
> This story is a blow for PostgreSQL > > http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&source=cxtnl&userid=aspire420@hotpop.com > thoughts ? Excellent news. The whole article is centred around a company moving away from PostgreSQL to a vendor which they have not chosen at this point. A whole article! It's not an article about a company choosing to DB2 or MSSQL with a mention that they were using "an Opensource database" or even a final comment stating that they have used PostgreSQL in the past -- but a whole article around a company moving away from PostgreSQL. You'll notice they mention "new database platform", which the author of the piece implies will include both hardware and software. I bet Oracle on a Quad Opteron and 8GB ram could beat the pants of PostgreSQL on a Dual Xeon with 1GB ram ;) Their qmail based mail solution is on a single Dell 1750 -- which also deserved a mention for some reason. Welcome to the big leagues -- we're making headlines on fluff pieces.
Vishal, > This story is a blow for PostgreSQL > > http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ >ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&s >ource=cxtnl&userid=aspire420@hotpop.com Can you or someone else in India contact them? Shridhar? The do not have an e-mail address for contact, just phone. http://www.skypakindia.com/location.htm I suspect that was the problem with their use of PostgreSQL. If they're not inclined to use the internet much, then they'd be totally unaware of where to get help. -- Josh Berkus Aglio Database Solutions San Francisco
Folks, Aha! I found e-mail for the company. Sent to their VP of IT. A copy of the message is below my .sig. HOWEVER, please do be aware that there is likely a great deal more to the story than the article presents; Skypak may be thinking of switching away from PostgreSQL for reasons that have nothing to do with actual performance. For example, they may be using it on Cygwin; you'll notice that the article doesn't mention a platform. -- Josh Berkus Aglio Database Solutions San Francisco ================================= Mr. Warnekar, We in the postgresql community were distrubed to read of your company's rather public repudation of PostgreSQL. Much more so because nobody in our active online community of developers -- including a number of Indians -- was aware that you were using PostgreSQL, let alone that you had problems with it. Was your staff aware that a great deal of expert, peer-to-peer help for PostgreSQL is available online? Were your problems raised on the PGSQL-PERFORMANCE e-mailing list or elsewhere? I spend quite a bit of time tuning PostgreSQL professionally for clients in the United States; it is how I support my participation in the Project. I find it somewhat hard to believe that PostgreSQL would not perform well under the loads you descibe in the article. Please consider sharing your issues with the community; I'm sure it would be cheaper to solve your performance problems with PostgreSQL than to port to another system. I look forward to hearing from you. -- Josh Berkus PostgreSQL Core Team http://advocacy.postgresql.org San Francisco
Re: PostgreSQL giving jitters to Skypak
From
"V i s h a l Kashyap @ [Sai Hertz And Control Systems]"
Date:
Dear Josh , >>http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ >>ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&s >>ource=cxtnl&userid=aspire420@hotpop.com >> >> > >Can you or someone else in India contact them? Shridhar? > > > I would do this and following is the list of questions I would be asking on the same time I would request others to add more , 1. Vacuum usage 2. postgresql.conf settings used (I am not a pro in this field) 3. OS used its not clear if PostgreSQL is on non windows platform. 4. I myself am not aware of any consultants in India that can provide PostgreSQL optimization. Others are requested to add more questions I would be calling on 27 August 2004, on the following number 22-28508181 Best Regards, Vishal Kashyap
Vishal, > I would do this and following is the list of questions I would be asking > on the same time I would request others to add more , > > 1. Vacuum usage > 2. postgresql.conf settings used (I am not a pro in this field) > 3. OS used its not clear if PostgreSQL is on non windows platform. > 4. I myself am not aware of any consultants in India that can provide > PostgreSQL optimization. > > Others are requested to add more questions I would be calling on 27 > August 2004, on the following number > 22-28508181 This all sounds good. Let's wait a couple of days to see how they respond to the e-mail. -- Josh Berkus Aglio Database Solutions San Francisco
Vishal, OK, looks like you should contact them by phone. The e-mail addresses on their web site are apparently out of date. Please read the letter I wrote as a guidline. I'd actually prefer that you call them tommorrow, if possible. See if you can reach Shridhar as well. If they are unresponsive, I will need to contact other publications to do some damage control, and sooner is better than later. -- Josh Berkus Aglio Database Solutions San Francisco
On Tue, 2004-08-24 at 11:52 -0700, Josh Berkus wrote: > Vishal, > > OK, looks like you should contact them by phone. The e-mail addresses on > their web site are apparently out of date. > Well, service@skypakindia.com seems to be their advertised contact address. It's the address for customer service as well as signing up as a reseller. A message sent there, addressed to the manager of IT or Suresh Chaudhary, would likely be forwarded to the proper person. -Steve
Hi ! I think it would be a good idea to check how they are using indexes, and what the structure of the database is. Missing an index or two can affect quite much. Rgs, Jussi > Dear Josh , > > >>http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ > >>ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&s > >>ource=cxtnl&userid=aspire420@hotpop.com > >> > >> > > > >Can you or someone else in India contact them? Shridhar? > > > > > > > I would do this and following is the list of questions I would be asking > on the same time I would request others to add more , > > 1. Vacuum usage > 2. postgresql.conf settings used (I am not a pro in this field) > 3. OS used its not clear if PostgreSQL is on non windows platform. > 4. I myself am not aware of any consultants in India that can provide > PostgreSQL optimization. > > Others are requested to add more questions I would be calling on 27 > August 2004, on the following number > 22-28508181 > > Best Regards, > Vishal Kashyap > > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org
Rod Taylor wrote: > Welcome to the big leagues -- we're making headlines on fluff pieces. Yeah, and evidently floating rumours that you're dropping PostgreSQL will immediately get you free help. :) -- Peter Eisentraut http://developer.postgresql.org/~petere/
On Tue, 24 Aug 2004, Jussi.Mikkola@bonware.com wrote: > Hi ! > > I think it would be a good idea to check how they are using indexes, and > what the structure of the database is. Missing an index or two can > affect quite much. My interest is in the type of hard drive interface, the drive(s) capacity and utilization and the size of the database. There is no mention in the article and though we on the list know better than to use IDE in (database) servers I've seen plenty of systems from major manufacturers come with IDE. Rod -- "Open Source Software - You usually get more than you pay for..." "Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL"
On Tue, 24 Aug 2004, Jussi.Mikkola@bonware.com wrote: > Hi ! > > I think it would be a good idea to check how they are using indexes, and what > the structure of the database is. Missing an index or two can affect quite > much. As I'm experiencing with a client right now ... schema hurts alot too ... all of their queries are bigint = int, so they are having to go through their code and changing it to bigint = int::bigint so that indices are being used properly ... > > Rgs, > > Jussi > > >> Dear Josh , >> >> >>http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ >> >>> ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&s >> >>ource=cxtnl&userid=aspire420@hotpop.com >> >> >> >> > >> >Can you or someone else in India contact them? Shridhar? >> > >> > > >> I would do this and following is the list of questions I would be asking on >> the same time I would request others to add more , >> >> 1. Vacuum usage >> 2. postgresql.conf settings used (I am not a pro in this field) >> 3. OS used its not clear if PostgreSQL is on non windows platform. >> 4. I myself am not aware of any consultants in India that can provide >> PostgreSQL optimization. >> >> Others are requested to add more questions I would be calling on 27 August >> 2004, on the following number >> 22-28508181 >> >> Best Regards, >> Vishal Kashyap >> >> >> ---------------------------(end of broadcast)--------------------------- >> TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org > > > ---------------------------(end of broadcast)--------------------------- > TIP 1: subscribe and unsubscribe commands go to majordomo@postgresql.org > ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
On Tue, 24 Aug 2004 20:11:13 -0300 (ADT), Marc G. Fournier <scrappy@postgresql.org> wrote: > On Tue, 24 Aug 2004, Jussi.Mikkola@bonware.com wrote: > > > Hi ! > > > > I think it would be a good idea to check how they are using indexes, and what > > the structure of the database is. Missing an index or two can affect quite > > much. > > As I'm experiencing with a client right now ... schema hurts alot too ... > all of their queries are bigint = int, so they are having to go through > their code and changing it to bigint = int::bigint so that indices are > being used properly ... I've always wondered... is there some reason we don't do "type promotion" to match indices? So if someone provides an int, and a bigint index exists, it should be used automatically, as they're interchangable (i.e. int is a subset of bigint). Chris -- | Christopher Petrilli | petrilli@gmail.com
Chris, > I've always wondered... is there some reason we don't do "type > promotion" to match indices? So if someone provides an int, and a > bigint index exists, it should be used automatically, as they're > interchangable (i.e. int is a subset of bigint). Because it's actually more difficult that it sounds. However, we *have* made changes to 8.0 to make it better at selecting appropriate types for constants when and index exists. -- Josh Berkus Aglio Database Solutions San Francisco
On Tue, 24 Aug 2004, Christopher Petrilli wrote: > On Tue, 24 Aug 2004 20:11:13 -0300 (ADT), Marc G. Fournier > <scrappy@postgresql.org> wrote: >> On Tue, 24 Aug 2004, Jussi.Mikkola@bonware.com wrote: >> >>> Hi ! >>> >>> I think it would be a good idea to check how they are using indexes, and what >>> the structure of the database is. Missing an index or two can affect quite >>> much. >> >> As I'm experiencing with a client right now ... schema hurts alot too ... >> all of their queries are bigint = int, so they are having to go through >> their code and changing it to bigint = int::bigint so that indices are >> being used properly ... > > I've always wondered... is there some reason we don't do "type > promotion" to match indices? So if someone provides an int, and a > bigint index exists, it should be used automatically, as they're > interchangable (i.e. int is a subset of bigint). This is fixed (or, partially addressed) in 8.0 ... ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
Re: PostgreSQL giving jitters to Skypak
From
"V i s h a l Kashyap @ [Sai Hertz And Control Systems]"
Date:
Dear All, > > This story is a blow for PostgreSQL > > http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&source=cxtnl&userid=aspire420@hotpop.com > As promised following is the list of questions I am going to ask the tech team of Skypak 1. Version PostgreSQL 2. OS and it's version. 3. Indexes, and what the structure of the database is. Missing an index or two can affect. 4. Type of hard drive interface, the drive(s) capacity and utilization and the size of the database. Not only this I will request them to post their queries to mailing list. Some apprihensions I have 1. I doubt they have any tech team. 2. Their website as per the Whois records is also not managed by them 3. If they do have a tech team they are M$ savvy as is ver much relevant from their website headers. -- With Best Regards, Vishal Kashyap. Yahoo IM : coeb_college@yahoo.com Director / Lead Software Developer, Sai Hertz And Control Systems Pvt Ltd. http://saihertz.com
On Thu, 2004-08-26 at 06:53 +0530, V i s h a l Kashyap @ [Sai Hertz And Control Systems] wrote: > Dear All, > > > > > This story is a blow for PostgreSQL > > > > http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&source=cxtnl&userid=aspire420@hotpop.com > > I have not seen any further activity on this. I was expecting to see something either here or on PERFORM by now. Just wondering if I missed something. It would be nice if this story had a happy ending due to "Online Assistance from PostgreSQL Experts Across the Globe" or some such. ;-) -Steve Bergman
On Thursday 02 Sep 2004 9:08 pm, Steve Bergman wrote: > On Thu, 2004-08-26 at 06:53 +0530, V i s h a l Kashyap @ [Sai Hertz And > > Control Systems] wrote: > > Dear All, > > > > > This story is a blow for PostgreSQL > > > > > > http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp& > > >section_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Bu > > >siness&source=cxtnl&userid=aspire420@hotpop.com > > I have not seen any further activity on this. I was expecting to see > something either here or on PERFORM by now. Well, we gave them really basic suggestions on performance tuning and they seem to have gone well. > Just wondering if I missed something. No, not much. Vishal is in touch with the concerned person. The person was unavailable early part of this week. So we do not have concrete results of tuning as yet. Besides I understand that the company is doing many other things in addition to posgresql tuning, with a limited manpower. So things are bit slow to trickle. > It would be nice if this story had a happy ending due to "Online > Assistance from PostgreSQL Experts Across the Globe" or some such. ;-) Definitely it would be that way. However it will take some time to reach it there. Certainly we do not want to show that we offered to help but have not much results to show, Do we? Of course, We will persue them further to get it done ASAP. Shridhar
On Thu, 2004-09-02 at 21:27 +0530, Shridhar Daithankar wrote: > > Definitely it would be that way. However it will take some time to reach it > there. Certainly we do not want to show that we offered to help but have not > much results to show, Do we? > Thanks. Sounds like things are going as well as can be expected. -Steve Bergman
Re: PostgreSQL giving jitters to Skypak
From
"V i s h a l Kashyap @ [Sai Hertz And Control Systems]"
Date:
Steve >>>http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&source=cxtnl&userid=aspire420@hotpop.com >>> >>> >>> > >I have not seen any further activity on this. I was expecting to see >something either here or on PERFORM by now. > > > I had telecon with the responsible person at Skypak . They are still with PostgreSQL and are using servises from Shridhar , Josh to solve their some problems. >It would be nice if this story had a happy ending due to "Online >Assistance from PostgreSQL Experts Across the Globe" or some such. ;-) > It is more of a happy ending we are pushing them to give a follow up experience to some other respectable publication, In india. -- With Best Regards, Vishal Kashyap. Yahoo IM : coeb_college@yahoo.com Director / Lead Software Developer, Sai Hertz And Control Systems Pvt Ltd. http://saihertz.com
On Fri, 2004-09-03 at 20:57, V i s h a l Kashyap @ [Sai Hertz And Control Systems] wrote: > Steve > > >>>http://www.cxotoday.com/cxo/jsp/index.jsp?log=false&file=template1.jsp§ion_code=1§ion=News&storyid=1383&subsection_code=1&subsection=Business&source=cxtnl&userid=aspire420@hotpop.com > >>> > >>> > >>> > > > >I have not seen any further activity on this. I was expecting to see > >something either here or on PERFORM by now. > > > > > > > I had telecon with the responsible person at Skypak . They are still > with PostgreSQL and are using servises from > Shridhar , Josh to solve their some problems. > > >It would be nice if this story had a happy ending due to "Online > >Assistance from PostgreSQL Experts Across the Globe" or some such. ;-) > > > It is more of a happy ending we are pushing them to give a follow up > experience to some other respectable publication, In india. Any chance we can get cxotoday to post a followup article about how the postgresql community contacted the company and helped them fix their problems?
On Sat, 2004-09-04 at 02:19, Shridhar Daithankar wrote: > On Saturday 04 Sep 2004 10:16 am, Scott Marlowe wrote: > > Any chance we can get cxotoday to post a followup article about how the > > postgresql community contacted the company and helped them fix their > > problems? > > We have reasons to believe that some other publication would be a better > choice. > > I tried contacting PC Quest, India but they have not responded in last three > days. I need to follow up bit more. I believe no other publication could be better since a) no other publication would have the same impact as the original one that posted the "hit piece" and b) the same audience may not be reached by another publication. However, I'm aware of the other possible arguments against publishing in cxotoday. Specifically, the original piece came across as a hackish attack with little or no research done on the part of the author other than to overhear a lunchtime conversation about how PostgreSQL just wasn't cutting it somewhere. However, it may be that this experience, and the telling of it to said author would be an eye opening experience for him (or her, I didn't really check.) I certainly believe that whoever publishes this story should reference the original one, and the incredibly professional way the postgresql community has handled it.
On Saturday 04 Sep 2004 10:16 am, Scott Marlowe wrote: > Any chance we can get cxotoday to post a followup article about how the > postgresql community contacted the company and helped them fix their > problems? We have reasons to believe that some other publication would be a better choice. I tried contacting PC Quest, India but they have not responded in last three days. I need to follow up bit more. Shridhar
On Sat, 4 Sep 2004, Scott Marlowe wrote: > On Sat, 2004-09-04 at 02:19, Shridhar Daithankar wrote: >> On Saturday 04 Sep 2004 10:16 am, Scott Marlowe wrote: >>> Any chance we can get cxotoday to post a followup article about how the >>> postgresql community contacted the company and helped them fix their >>> problems? >> >> We have reasons to believe that some other publication would be a better >> choice. >> >> I tried contacting PC Quest, India but they have not responded in last three >> days. I need to follow up bit more. > > I believe no other publication could be better since a) no other > publication would have the same impact as the original one that posted > the "hit piece" and b) the same audience may not be reached by another > publication. However, I'm aware of the other possible arguments against > publishing in cxotoday. Specifically, the original piece came across as > a hackish attack with little or no research done on the part of the > author other than to overhear a lunchtime conversation about how > PostgreSQL just wasn't cutting it somewhere. However, it may be that > this experience, and the telling of it to said author would be an eye > opening experience for him (or her, I didn't really check.) > > I certainly believe that whoever publishes this story should reference > the original one, and the incredibly professional way the postgresql > community has handled it. Stupid question, but could this backfire on us? Here is a company in India that complained about the atrocious performance PostgreSQL was giving them, they complained vocally about it, and "the community" jumped to the rescue? Are we going to jump the rescue of the next one too? Are we setting a potentially bad precedent by advertising that "if you yell loud enough, we'll focus our attentions on you"? ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
Scott, Marc, > I believe no other publication could be better since a) no other > publication would have the same impact as the original one that posted > the "hit piece" and b) the same audience may not be reached by another > publication. As you guessed, we have already contacted CXOToday, and researched the background of both the publication and the reporter (thanks Simon!). Based on both research and our interaction with the reporter, we do not believe that any story from us would receive fair treatement. > Here is a company in India that complained about the atrocious performance > PostgreSQL was giving them, they complained vocally about it, and "the > community" jumped to the rescue? Are we going to jump the rescue of the > next one too? Are we setting a potentially bad precedent by advertising > that "if you yell loud enough, we'll focus our attentions on you"? We won't be giving them any help they couldn't otherwise get simply by subscribing to PGSQL-PERFORMANCE or jumping on IRC. The reason that Skypak ended up in trouble is that they are NOT interent-saavy at all; their installation of PostgreSQL was put in by an apparently less-than-competent constulant who then disappeared when they ran into problems (at least, according to Skypak). So they were completely unaware of the existence of postgresql.org, the mailing lists, or online documentation. I, personally, don't have any problem giving them some help in getting connected with these resources -- the same as I do for about 10 people a day who contact me through advocacy.postgresql.org or press@postgresql.org. Also, I think I can get a good news story out of it, which is worth a little effort. -- Josh Berkus Aglio Database Solutions San Francisco
On Sat, 4 Sep 2004, Josh Berkus wrote: > Scott, Marc, > >> I believe no other publication could be better since a) no other >> publication would have the same impact as the original one that posted >> the "hit piece" and b) the same audience may not be reached by another >> publication. > > As you guessed, we have already contacted CXOToday, and researched the > background of both the publication and the reporter (thanks Simon!). Based > on both research and our interaction with the reporter, we do not believe > that any story from us would receive fair treatement. > >> Here is a company in India that complained about the atrocious performance >> PostgreSQL was giving them, they complained vocally about it, and "the >> community" jumped to the rescue? Are we going to jump the rescue of the >> next one too? Are we setting a potentially bad precedent by advertising >> that "if you yell loud enough, we'll focus our attentions on you"? > > We won't be giving them any help they couldn't otherwise get simply by > subscribing to PGSQL-PERFORMANCE or jumping on IRC. > > The reason that Skypak ended up in trouble is that they are NOT interent-saavy > at all; their installation of PostgreSQL was put in by an apparently > less-than-competent constulant who then disappeared when they ran into > problems (at least, according to Skypak). So they were completely unaware > of the existence of postgresql.org, the mailing lists, or online > documentation. I, personally, don't have any problem giving them some help > in getting connected with these resources -- the same as I do for about 10 > people a day who contact me through advocacy.postgresql.org or > press@postgresql.org. > > Also, I think I can get a good news story out of it, which is worth a little > effort. Agreed ... the only concern I have is the way that they went about getting that help ... the whole 'pre-article' and all that ... the whole thing just has a feel of 'if I cry loud enough, someone might come along and fix it for me' ... its the public denounciation, followed by us rushing to his aid, that puts the hairs up on the back of my neck *shrug* ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
On 4 Sep 2004 at 11:49, Josh Berkus wrote: > As you guessed, we have already contacted CXOToday, and researched the > background of both the publication and the reporter (thanks Simon!). Based > on both research and our interaction with the reporter, we do not believe > that any story from us would receive fair treatement. That conclusion is supported by the opening sentences in the article in question. I got the impression they were anti open source. -- Dan Langille : http://www.langille.org/
On Sat, 2004-09-04 at 16:03 -0300, Marc G. Fournier wrote: > Agreed ... the only concern I have is the way that they went about getting > that help ... the whole 'pre-article' and all that ... the whole thing > just has a feel of 'if I cry loud enough, someone might come along and fix > it for me' ... its the public denounciation, followed by us rushing to his > aid, that puts the hairs up on the back of my neck *shrug* > If they had come to the lists with their problems, not solved them, and then denounced loudly, I'd agree. In this case it sounds like a consultant set it up, didn't do any tuning and left, and they were left with something that wasn't working well for them and were frustrated, and possibly angry. Different situation, really. Fortunately, they are getting assistance and hopefully they'll come away happy and sing praises to the world. ;-) What bothers me more is the idea of having an actively PostgreSQL- hostile journalist at CXOToday. -Steve Bergman
Steve, > What bothers me more is the idea of having an actively PostgreSQL- > hostile journalist at CXOToday. I wouldn't worry about it. CXOToday is probably India's least popular online tech magazine. Witness how NO other news services have picked up the story. As we get more successful, we can expect to make "enemies". In that regard, we're moving in the right direction; we've been mending fences with MySQL and are starting to draw heat from Oracle and DB2 (and I'll bet that MS will trash us after the Windows release). These are the people we want to see angry with us in print because it means that we're a threat to their business. -- Josh Berkus Aglio Database Solutions San Francisco