NTT America announced today that it has leased a new 15,000 square foot data center in Santa Clara, CA. Big deal, you might think, a network provider expands its capacity. Except this network provider has a very high-profile client: Twitter.
Though Twitter is never actually mentioned in NTT’s press release, the blog Data Center Knowledge put two and two together, recalling a quote in June from NTT America COO Kazuhiro Gomi: “traffic generated by Twitter is getting so big, it’s basically eating up a lot of our data center network resources, especially the segment where Twitter is hosted. Other customers are riding on the same segment.”
NTT America’s name came up quite a bit recently during the DDoS attacks that crippled Twitter. As Twitter’s network partner, NTT put in place many of the safeguards that slowed the attack enough so that Twitter could get the service back up. You may recall that Twitter also worked closely with NTT to reschedule a planned maintenance in June to make sure that the Iranian protest messages could continue to flow over the service.
Eventually, if Twitter continues its rapid growth, you’d think they would want their own data centers, like Google and Facebook have. But the leasing of this large new data center would seem to indicate that Twitter will be onboard with them for the foreseeable future. Twitter has been with NTT America since early 2008, following widespread reliability issues.
The idea of Twitter going down in the event of a major earthquake in the San Francisco Bay Area is occasionally brought up when the question of Twitter as a reliable form of communication is raised. NTT America says this new data center “meets seismic zone four specifications.” NTT also notes that the data center, “is equipped with redundant power feeds, and carrier-class uninterruptable and back-up power. Optimal systems performance is maintained by fully redundant water cooling systems coupled with advanced humidity and temperature controls. Continuous monitoring ensures that all systems are fully operational.”
Could this really mean the end of the Fail Whale?
Last time I checked they didn’t have any ads or anything similar, where is the money for all of this coming from ?
who? Twitter or NTT?
Twitter of course.
well they’re not paying for this datacenter, NTT America is. though they are paying NTT, and they have plenty of funding for that
The common issue discussed is Twitter and NTT were partners for long time and Twitter has acted as a client for the NTT. The combination of both worked very well in DDOS attacks and Iranian issue.
But, the investors are pitty here.Even if there is screening about Twitters failwhale between both,it is quite natural.
Both are very good organizations and they can work independently..There comes a good oppurtunity for Twitter to open their own Datacenter eventually?
Investors.
AKA suckers.
twitter have more money than they have fail whales.
Now that, is a graph i would like to see…
Ah, thought so. I think all parties involved are getting very serious about this. Twitter didn’t break the 99% uptime floor for August either
http://tinyurl.com/uptimeFTW
After this move I’d expect Twitter to stay well north of 99.5% if not 99.95% like they hit back in March 2009.
yep, this is good stuff. you probably have some good thoughts on Twitter going to its own datacenter eventually?
I’m sure Twitter is being called on by the likes of @drtdatacenters and others in high touch solution services as the portability and spread of Twitter is a given eventuality. What people see when they update Twitter is never the whole of the business — the back side views, feed(firehose), and making that happen for partners/etc. has to be something you can count on if you build to it. So, control, ownership, and a immediate response time need will mandate expansion.
A likely emphasis might be placed on the early Facebook-like concerns that are unique to services that are gaining popularity — latency. While NTT -might- be able to respond, it’s often the case that an early hosting relationship will change in favor of a better fit later that maps to geographic requirements.
You’d think they’d have load balancing across two or more datacenters by now.
Yeah, an internet business relying on a single point of failure? wat the fux.
+1
its not really that hard to set up either
-1
It is if you’re Twitter
True, you’d have to remember the root password to your server….
That’s no big deal, all you need is a French hacker/cracker to remind you.
Twitter uses amazon s3 for the images and so there is very little bandwidth to worry about if you think about it. Twitter is basically one huge mySQL database. This news is more of the same PR nonsense.
does anyone really think datacenter capability has anything to do with uptime for sites like Twitter?
Twitter’s problems are almost certainly all application and ops problems (bugs, human error and scalability issues in their app) none of that is data center related at all. Of course if you’re in a crappy location and they dont maintain their UPS you can have problems, but as a % of total outages, datacenter related issues are very small.
They could have the biggest datacenter in the world and still could and absolutley will have outages, guaranteed.
As for using NTT to help with their DDOS attack.. they just need more equipment (money and time) and more/better network guys (money and time) is all; that will come as ops matures (give em a break; they scaled up faster than the ops team can handle, that’s the price of success of course.) Once they’re scaled up enough they wont need NTT to handle this kind of thing anymore just as FB handled the attack without much issue.
btw.. 15,000 sqft is not a big datacenter, quite the opposite actually, it’s tiny, so tiny that I would think NTT are leasing space inside a larger facility and reselling (to people like Twitter)
All in all.. a none story. Move along, nothing to see here.
does anyone really think datacenter capability has anything to do with uptime for sites like Twitter?
Yeah, actually.
There are two things I can imagine a new datacenter helping with. The most important one is power. This is the single biggest thing limiting datacenters from expanding capacity in place. The other one is additional IP carrier capacity. If they have different carriers there, or more capacity, that could obviously help, although it’s not nearly as big a deal as power.
However, marketing blurbs aside, this datacenter is just as vulnerable to a large quake as any other datacenter in Northern California. Twitter is going to have to go active/active in two geographically distinct datacenters, preferrably in two different legal jurisdictions, before you could give them any kind of positive review in terms of business continuity planning.
Yes. Capacity can matter quite a bit. It’s not just about ups and techs. Capacity plays a huge role. There is a reason why tier I access to dark fiber is sought among other things. Also, NTT is listed as on of the worlds largest companies. When NTT got into the internetbusiness they bought the worlds largest Internet company. They are huge. So no. They are not sub leasing space as a third party. They put up dedicated satalite data centers everywhere. This is done fordfar more reasons than I could list xomfortly here.
Thank God. I was worried that the daily onslaught of hookers and snake oil salesmen following people on twitter might be slowed down.