![frisk design : building websites that work [link to homepage ]](/images/gui/logo6.png){kind=logo}
Jan
29
2008
Wordpress, slow, Mediatemple
These were the terms I typed into Google tonight to try and get some clue about my slow performing Wordpress blogs. The results that came back were enlightening: far from being the reputable company I once thought, MediaTemple have become a bit of laughing stock. Why? Overselling.
Mediatemple have been well respected for years for offering well priced, high performance hosting. Many big names in the web development world use them and coupled with the promises of their new Grid Server package, there was much to feel good about. So I signed up with them last July and got my own plot on a GS which promised all sorts of great features, including bandwidth rollover.
It turns out that their technology simply couldn’t cope and while they did spend time and money on trying to fix the issues, they also did a good job of blaming their customers. Although performance wasn’t too bad to start with for my own websites, in the last couple of months it’s got steadily worse.
It was the last straw tonight. The reason that I haven’t updated this blog for a while is because of the slowness of using Wordpress. I guess it’s fitting that my first post in a while should explain why. This evening I timed the page loading time of my Wordpress Write Post page using the great Firebug and it was over two minutes! In this day and age, especially on my 20Mb broadband package, that is insane — and more importantly, unsustainable.
I’m running a web design business, not simply blogging for fun, so I find it completely unacceptable to have these kinds of issues on what is supposed to be one of the best hosting solutions in the business. I’ve emailed their support asking for the matter to be looked into and resolved, so I guess I’ll have to wait and see what they say.
I don’t have much confidence though. The last time I emailed support, it took them five days to answer and that was over me accidentally locking myself out of my admin panel. Five days to reset my password! Would you put up with that from your web host?
A rhetorical question.
If I don’t get a satisfactory outcome to my support request and if things don’t improve, I shall once again be in the frustrating position of trying to find a reliable, reasonably priced and effective host. Wish me luck. Or better still, please recommend one…
Mark Zakariya
29/1/2008 11:29pm
I know what you’re experiencing, in regards to incompetent support and lengthy load times. Often times I ask for something very specific, and they give me a generic copy/paste answer. It’s to the point now where I specifically ask in a support request for a “real” answer, from someone there who knows what I’m actually asking.
The load times have their ups and downs. Sometimes my blog loads at a “normal” speed, and sometimes it just is so slow I get sick of looking at my own site. Despite various phone calls, emails, support requests, and basically begging for an answer, I got nothing- until recently.
They are creating (yet another) beta product that will supposedly fix the issue. The (cs) Cluster-Server. It hasn’t been released as a beta yet, but they signed me up for it.
If it doesn’t help (which wouldn’t surprise me, since I specifically noted speed as in issue in the GS beta), I’ll simply move all my sites over to A Small Orange, which has an excellent reputation amongst webmasters.
Best of luck,
Mark Zakariya
Alex @ (mt)
29/1/2008 11:35pm
Matthew,
I understand your frustrated and I’ve forwarded your post onto our support team…We’ve been seeing issues with UK to USA hops and have had several clients do “traceroutes” to see where the lag resides. I’d like to show you that pingdom is showing your page loading at 5.1 seconds: http://tools.pingdom.com/fpt/?url=www.friskdesign.com&treeview=0&column=objectID&order=1&type=0&save=true
and someone discovered a BT issue in Dec. we’re intially blamed for:
http://ask.metafilter.com/77991/Is-BT-blocking-sites-on-me
As for database issues, currently, it’s nothing like what we discussed in Jan ‘07 (our blog post you linked to) but we have had some down time and it’s still not up to our high standards.
Someone from our team will respond to your ticket and we’ll investigate if it is us or an ISP. If you don’t hear from someone shortly, please email me and I’ll make sure they do. I want you to know whats happening (either way) and you can make the best decision for the sake of your business.
Cheers,
Alex @ (mt)
30/1/2008 12:12am
I forgot to add that 5 days for a response is not acceptable. I’m looking over your ticket history and I’ve asked our head of support for a reason why this happened.
Cheers,
Matthew Hill [friskdesign]
30/1/2008 12:21am
@Mark: Thanks for the heads up about ASO — I’ve heard good things about them before. Of course, if they are reliable currently because they are small and not well known, will they be able to upscale to cater for disgruntled customers leaving other hosts? ;-)
@Alex @ (mt): I’m new to blogging and forget the power of the ping and the trackback, so I was surprised and delighted to receive your reply. Thank you for taking the time to respond. To be fair, the five days I mentioned was over a weekend and I don’t know your weekend support policy — plus, because I’d locked my self out of admin, I had to send the support request to a standard enquiry email on your site. Neverthless, I still found it a frustrating experience having to wait so long for a resolution.
Bernard Hatch
30/1/2008 12:28am
Alex,
I have to say you guys are really pulling out all the stops to try and blame someone else. We are currently experiencing the exact same issue, and have been for some time. Take a look at our pingdom results = 26.7 secs. Thats ridiculous.
http://tools.pingdom.com/fpt/?url=www.tierrastudios.com/crm&treeview=0&column=objectID&order=1&type=0&save=true
Please just do the right thing and admit to your customers that this is a problem with your service and offer us some kind of reasonable solution. ie. help migrating to your far more reliable dedicated virtual server. We have been patient and have done our part to investigate any and every possibility on our side. You guys sent us down every avenue possible and still haven’t given me any date for resolution for this issue. When I called tech support today, to inquire about getting signed up for the cs beta or for help migrating our sites to your dv service, I was told that any and all migration work would have to be done by our team and that we were ineligible for the cs beta because we were not “pushing the limits” of our current server.
I have been with your company for quite sometime and we operate several websites and businesses using your service. I am appalled that you guys have not given us any help or explanation of possible solutions. Just a sorry I can only relay information.
Do the right thing, if you’re having problems with the grid server, then help us move to a more reliable and effective service. Don’t keep telling us that its our problem.
Bernard Hatch
30/1/2008 12:43am
I would like to add as a side comment. I have been asking, practically begging for help with this issue. We have no problem taking on any workload to fix a problem that has resulted from a configuration issue, or from un-optimized code. However, your tech support personnel have been unable to help me pin-point where that problem might lie (we even tried trouble shooting on a completely fresh, unaltered version of wordpress, still slow).
Arman @ (mt)
30/1/2008 5:45am
Hello all,
I’ve been working with Alex and our tech support team on this slowness issue. We do know about the latency and published this notice earlier today: http://weblog.mediatemple.net/weblog/category/system-incidents/gs-grid-service-cluster2-web-latency-01-29-2008/
I would like to help clarify a few things:
Re: Shifting the blame - This is not what we’re doing. Many factors can affect the way a website performs. Some of the time, slow performance really _is_ caused by inefficient coding, so it’s best to start there when troubleshooting an issue. Other times, it is legitimately caused by something on the host’s end. If you read the page I linked above you’ll see that we are not trying to blame someone else for these problems and are clearly communicating what is being done to make things better.
Re: No (cs) for not “pushing the limits” - I really think this was miscommunicated somehow. The (cs) Cluster-Server is not currently in beta testing (but will be soon). I don’t know exactly how your talk with the Agent played out, but that was the wrong answer. It is possible the Agent thought you might have been referring to (cx), which is the abbreviation of our (cx) Complex-Hosting product. I really can’t say for certain, but our support department has never been instructed to give such an answer to that question. Whatever the reason, we apologize for the misunderstanding.
Best,
Arman Zakaryan
(mt) Media Temple, Inc.
Bernard Hatch
30/1/2008 5:20pm
@ Arman@mt: I appreciate the response and the blog posting discussing the issue. I have just been on the receiving end of several “it’s code optimization” discussions. It has just been a long road to get anyone to admit to us that this not our issue. I began calling about our site’s performance in dec!
Re: the cs beta. I don’t believe there was confusion regarding the service types. I specifically asked about the beta. Your tech told me it was not released yet, and would only be released to “site admins and other customers who were pushing the limits”. I believe he was just trying to warn me that it was not going to be a completely safe sandbox, but I really just didn’t understand his logic. If we’re calling about issues related to performance on our sites and we have gone through the steps to try and optimize things on our end I believe that makes us perfect candidates for a beta program. Especially if that service is designed to eliminate the issues we are currently having. It seems odd to me that he would tell me the cluster server was designed to address the problems we are having, but not offer it as possible shorter term solution if we were willing to shoulder some of the risk by being included in a beta test. Ultimately I don’t really care about being included in a beta, I just want our sites to function properly. I just saw it as an opportunity to help a previously reliable vendor and fix some of our issues. If there are other options for getting these things fixed I eagerly await them.
Anyway, I am pleased that your techs are looking into the issue.
Mark Zakariya
30/1/2008 8:15pm
Arman, Alex, we’ve been down this route before. Quite frankly, I’m not saying thanks (or anything, for that matter) until something happens on February 8 that completely changes everything. The only reason I’ve stayed so long is because you two have provided me with such outstanding support, and because the (gs) package is rather competitive (in regards to space) in comparison to ASO.
@Alex: If you are the one I spoke to two months ago on the phone .. what happened? You were courteous enough to warn me that you were very busy at the time, but then you vanished. We were discussing MySQL issues, as well as the (cs) Cluster-Server beta. Please get back to me when you get the chance.
@Arman: I’m glad to see you guys are doing something, or at least acknowledging something is wrong! You were *great* with the beta, and I hope you guys are the same way this time.
I don’t personally have an issue with (mt), but I host another site for a family financial business, and they often complain of *VERY* slow email downloads, as well as slow website loads. Business-class T1 doesn’t fail, so don’t bother pinning it on them.
Although, I’m really glad that this blog post has drawn some attention from (mt).. but I also notice that this isn’t the first time. I’ve seen quite a few blogs get replies from (mt) reps, with nothing done.
Best Regards,
Mark Zakariya
Bernard Hatch
30/1/2008 9:02pm
I would like to echo what Mark said above. The only reason I have stayed this long is your previously outstanding tech support. Hopefully this trust will be well founded and something significant will change on Feb. 6.
Dan
15/3/2008 8:36pm
Hi,
How are you finding (MT) these days? I have to say I feel a certain fondness towards them, but I’ve had my doubts about speed myself. As you said, the Wordpress Write page seems to take a very long time to load up.
Interested to find someone who shares the same view as me, and even more curious to see if your opinion has changed lately.
Matthew Hill [friskdesign]
19/3/2008 2:51pm
Hi Dan — sorry your posts didn’t appear straightaway, I was on holiday. Overall performance has definitely improved since the recent maintenance.
The Wordpress Write page can still be interminably slow though. As I only have experience of running WP on Mediatemple kit, I don’t know if it’s still a hardware issue or a problem with Wordpress. I’ve been experiencing a fair few issues with slow scripts on Wordpress recently, so it could be the software, but at this point I really wouldn’t like to say one way or the other.
Dan
19/3/2008 2:56pm
No problem, sorry I spammed your blog ;)
I’ve signed up for an account with a highly regarded “other” host to do some testing. First impressions are that MT is a lot slower than the “other”, from the UK at least.
Going to get some more of my sites moved across before I say for sure though. I’ll keep you posted.
Matthew Hill [friskdesign]
19/3/2008 6:18pm
Very interested to hear how you get on with your new host, Dan. Please do keep me updated, thank you. Interestingly, this is the third time I’ve tried to post this reply — I keep getting timeouts and the site has been disappearing randomly all day!
archondigital
25/3/2008 3:21am
I’m also on the (gs) and it keeps having too many incidents of downtime lately. Everyone’s talking about (cs) and I wonder if they’ll migrate existing (gs) clients automatically.
Right now the only thing holding me back from switching to slicehost is the learning curve for setting up and handling your own VPS but I’m getting there.
Dan
25/3/2008 1:54pm
I toyed with going with SliceHost, but can’t be bothered with keeping my “slice” up to date in terms of security and things.
I’m just finishing a move to WebFaction, and will post a review and comparison on my blog very shortly.
Matthew Hill [friskdesign]
25/3/2008 3:37pm
Thanks for the update Dan. I’ve never heard of Webfaction and googling for reviews of them throws up very little which worries me somewhat. But maybe that’s a case of people only review hosts when they’ve had bad service…. Can’t wait to hear your review of their service. I’m still experiencing slowdowns with MT and really want something a bit faster.
sam
26/3/2008 12:50am
Looking forward to hearing the feedback on webfaction…
Dan
27/3/2008 6:20pm
My thoughts on the MT versus WebFaction stuff can be found here: http://www.istherefood.com/2008/03/27/webfaction-versus-media-temple/
I’m going to post another piece on WebFaction’s control panel in the next couple of days.
Chakabrah
31/3/2008 9:03pm
“these systematically pathological applications overloaded every database we put them on – no matter how big or resourceful”
OH! You mean like Drupal 6?
tuneczar
02/4/2008 5:10am
Been with (mt) for several years, but moving our company site to A2 now… Although I appreciate the customer service attempts by (mt), we’re leaving due to too much sporadic down time. We want more control and responsiveness when issues occur, so that’s the main reason we’re moving away from (mt).
Michael
06/5/2008 9:48am
I understand your frustration. I have a static html site with them and a separate WP+MYSQL site. The WP site really, really drags, pushing 10-15 seconds to load a page, when it isn’t reporting downtime, which is typically at least 2 hrs a week. Some of the issue is plugins, but still, using the TamperData plugin to time which assets chew up page load time, (mt) is at least 100% slower than a direct test on a moderate loaded machine.
When I spoke with them, they claimed it was “my code”. Being that I am also IT Director for a couple of larger ecommerce sites, and had timed (mt) against my own local VM installed instance, I can definitely say that the performance diff I noted is pretty squarely a combined cpu & db responsiveness issue, as you say, overcapacity. Doesn’t make much sense to me, on our dedicated boxes we would have to be SERIOUSLY over capacity to have performance as poor as what we get from (mt). I am not impressed.
Our other gear is in Rackspace, and I have thought on a couple occasions of trying their Mosso package, which appears to be under a much better and proven management model.
Time to go back to netcraft.com and check the survey.
Michael
06/5/2008 10:04am
Some additional detail - http://jasonzhao.hotpng.com/2008/01/10/media-temple-grid-server-and-database-performance-review.html
and likely theory explaining it…
Ross
12/6/2008 2:08am
Seems like things may have gotten a bit better but I’m not holding my breath. How are things going for you?
Ross
- http://www.hostdisciple.com
Matthew Hill [friskdesign]
12/6/2008 9:14am
Hi Ross. Overall, performance is improved but I wouldn’t describe it as fast. Wordpress still takes an age to load and the speed of FTP is still pretty bad. I’m about to try out Webfaction to see how they fare.
Another reason why I wasn’t keen on staying with Mediatemple is that I don’t use credit cards and my only option to pay was through PayPal, but MT don’t offer a monthly option when paid this way, you have to pay for a year up front. I’m not keen on being locked in for another year to a service that could again not deliver on its promise of speed and reliability.
Ross
12/6/2008 2:47pm
Matt,
Webfaction is a great company, you may also want to check out www.medialayer.com
Ross
- http://www.hostdisciple.com
Ehab
21/6/2008 11:02pm
Thanks for the lead :) I thought I was alone with this !