LooseSuits

Hmm, even more interesting, the SmartPool v.2 crap they're putting non-container customers on is limited to 20 max connections. That's like nothing! Users like myself will be forced to upgrade to containers or risk having mySQL connection overages.

MediaTemple has a huge cult following; which means they price gouge whenever possible. It's the sad truth. They can call their servers whatever cool names they want, honestly they're damn good advertisers and they do this all well, but it all just comes down to business.

I just avoided them and went with a VPS server instead. It's much more stuff for your money - I recommend VPSlink.net. It's actually around the same price, and you have full control over your server.

Honestly, I am a little upset. Although it is understandable because without these caps and restrictions, grid servers such as mine (which I did complain about and get in an argument with (mt) themselves) would be and have been slow.

That's a bit frustrating, but I see people paying the high price. So far I've been satisfied with their Ruby on Rails container.

MediaTemple's old (ss) platform may not have up to date MySQL (no WordPress 2.1), but there's some huge advantages to procrastinating and not upgrading. No MySQL errors, no MySQL limits and a rather empty shared server. It was unbelievable how zippy WordPress was while on the Digg front page (as well as TUAW, Gizmodo, Wired & others), without the use of anything like wp-cache (I installed it but forgot to enable it). I've thought about upgrading to the (gs) so I can get my 100GB bandwidth back, but at this point there's no rush whatsoever. When they force me to upgrade, I will, though it would be nice if that doesn't come anytime soon.

This shouldn't be much of a surprise for anyone running rails on the gs.

When I had mine, I had a single copy of Mephisto, and a super simple site that interfaced with the Flickr API, and they shut me out until I bought an upgraded container.

I've since moved on to A Small Orange, where I'm running no less than that Flickr app, a copy of Mephisto, a copy of SimpleLog, and a handful of small projects I'm working on (which I can even run in development mode if I so wish no questions asked) all for less than what the gridserver cost me excluding the cost of the rails container.

Don't get me wrong, the people at (mt) were very nice and offered me several months of free hosting when I decided to leave, but moving on was the best choice I've made.

jbarket, are you on a shared account at ASO?

I switched to (gs) primarily to flee the "bad neighbor effect." Indeed, this is what (mt) claims the GridServer eliminates.

While MySQL SmartPool v.2 is laudable for its enhancements vis-a-vis traditional shared hosting, it is still susceptible to the "bad neighbor effect." To whit:

Even with the proprietary enhancements of the MySQL SmartPool system, it remains a "shared system" where one or more users may be competing for system resources. This can lead to a negative system condition and reduced performance. Although the SmartPool does an excellent job at combating the downsides of shared platforms, it does not completely guarantee 100% uptime or performance. This is a reality for current versions of MySQL Server in any shared system, from any hosting service provider.

... unless, of course, you're on the GRID. Right? That's the whole idea! That's why I got on-board!

To that end, I would like to see the Lite MySQL GridContainer offered as standard-issue for all GridServer accounts, keeping in line with the original promise of no "bad neighbor effect."

I have also written MediaTemple requesting the same.

Word.

After months and months of crappy service I would like to see them go the extra mile and have these containers be standard.

This is ridiculous given the MySQL issues every customer has been experienced.

On top of this, somehow, with traffic only increasing 5-10% in this billing cycle, my GPU usage has increased 150% from 450 to 1100. Across my sites (mostly WP blogs), I get about an avg of 1000-1500 visits per day, so this not huge traffic.

I emailed them about and three days later and got this

"As an initial bit of information, there was actually a bug in the way GPU's where being processed and accounted -- they were being counted below our published spec for about 6 weeks in the last months. That explains why, if your traffic is consistent, you would have seen a jump in usage."

So this proprietary blackbox statistic (GPU) was miscalcuated behind the curtain by then and now that they "fixed it" I am using more than twice as much?

I thought something like that was up. Thanks johnnydebacle for the information.

OK..enough is enough.

everyone read this page.
http://www.mediatemple.net/webhosting/gs/mysql-pool.htm
the container upgrades are for extremely high CPU usage.

the Smartpool more or less gives EVERYONE a free upgrade and allows you to burst, say if you get Dugg as well..if your MySQL usage stays at a constant high rate, you will need to upgrade.

Jason

To clarify, the concern isn't around usage, but rather the availability of MySQL under SmartPool.

Jdandrea...i'm not sure what you mean:

"When the SmartPool system recognizes sudden spikes in database traffic, it allocates a single-tenant MySQL Container which allows the database to grow beyond its guaranteed resources. This unique feature helps customers survive temporary traffic bursts resulting from press releases, mass blog coverage, DIGG, etc."

and

"Rather than making everyone share and compete for resources on a single MySQL database, the SmartPool distributes users amongst a farm of MySQL Container Servers. Using virtualization technology, the SmartPool achieves a low user density per MySQL instance, which minimizes resource competition and improves overall system stability."

please clarify what you mean by availability

You can review the product page here:
http://www.mediatemple.net/webhosting/gs/mysql-pool.htm

Jason

Jason,

See my earlier excerpt from the SmartPool page (the same page you have referenced). In particular:

Although the SmartPool does an excellent job at combating the downsides of shared platforms, it does not completely guarantee 100% uptime or performance.

To clarify, I'm referring to availability in terms of uptime.

SmartPool is terrific for what it does but it remains a shared system. There's the rub: GridServer is not advertised as a shared system, yet MySQL with SmartPool is standard-issue vs. the MySQL GridContainer.

My idea is that the (gs) gives priority to high-demanding websites. So if you have low traffic or very displaced traffic, your users are going to experience those very lame lags (an extremely simple PHP page can take up to 4000ms to be rendered, which becomes nearly 7000ms if mySQL is involved, even just to open the connection, without sending any query!!!!).

After months of nagging, they still are blaming my "bad coding". There is possibly no optimization issue, as I am only keeping 2 really, really simple projects in there (db hasn't yet been populated with real data: there are some 6 rows in each table).

____ edit:

I must admit I have experienced improvements by the day in the past week: probably about 50% reduction of times, which means they are working on it.
____ edit end

They have offered a 1 month of free mySQL Container Lite to analyze queries. Obviously everything I got was "no slow queries at this time".

I have uploaded a large recordset from a project I've worked on, currently running on an Italian dedicated server (along with other 50 sites). A table with some 67 thousand rows.

The shared mySQL would take 3700ms to perform a COUNT(*) until caching, then dropping to 188ms (which appears to be the lower limit).

In the container performance boost was quite amazing: 65% time reduction on first queries and over 95% time reduction once things get cached, this means less than 10ms, which is extremely good.

The problem is that 128MB is not much estate at all (dunno how much of it is actually available for operation), and I'm not willing to pay extra money to workaround the poor performance of what I'm already paying for.

mySQL Container has too much performance in too little space, while the (gs) is falling short of the specifications and expectations.

My (ss) was far better balanced, nothing was implessive, but everything was pretty consistent.

In less than two weeks I will have to find another host: my client is getting pissed.

@johnnydebacle my AccountCenter control panel doesn't show GPU usage at the moment: "This component is currently being upgraded to provide more accurate reporting. GPU accounting for your site has been temporarily disabled."

I've seen the offer over at VPSlink.com, prices look very aggressive.
Does anyone have experience with them?
Reliability, performance, customer care?

p.s. (mt) customer care has gone wild since (gs) was released. Before you'd get a response in max a few hours.. now it takes days, and most of the time it's the lame generic useless answers that you'd expect being given to newbies. Quite annoying.

Let me give you a great quote from (MT) support when my database was slow as... you know what - my WP pages were showing 30 second MySQL queries on average:

"How often are you running into this issue? Right this very second query times on that database seem to be fine. It could be that you where running into the infamous "bad neighbor" effect where one of the other users you "share" operating space with begins to consume an excessive amount of resources. One way to ensure this doesn't happen is to move to a Mysql container, separating yourself from the SmartPool as outlined in the knowledge base article below.

http://kb.mediatemple.net/article.php?id=536

If you have any further questions regarding your (mt) Media Temple services, please feel free to contact us at any time."

But when you're on the Order page you see this:

"Rather than making everyone share and compete for resources on a single MySQL database, the SmartPool distributes users amongst a farm of MySQL Container Servers. Using virtualization technology, the SmartPool achieves a low user density per MySQL instance, which minimizes resource competition and improves overall system stability."

I had thought on upgrading to a (dv) but I heard so much bad stuff about them and their support has been generally so lousy that I'm currently waiting for my slice over at slicehost.com and will be moving pretty soon now.

zattraz, I upgraded to a (dv) Extreme account from a (gs) account and it's never let me down. The (dv) product seems a lot more stable than the (gs).

Thanks for the info, bobbyh. I will probably be recommending that to one of my blue blood clients that gets ~10.000 pageviews/day and who needs something managed.

I was actually referring more along the lines of the (dv) base that needs a lot of tweaking and I dislike their UOP breaking on anything.

See http://davidseah.com/notebook/configuring-media-temple-dedicated-virtual-server-for-wordpress/ and http://davidseah.com/notebook/other-optimizations-for-the-media-temple-dv-base/