louisgray.com - Latest Comments in http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

John E. Bredehoft (Empoprises) — Mon, 11 May 2009 16:20:14 -0000

guruvan, you're right. Redundancy is good. Even if a service doesn't have a technical failure, it can always have a business failure.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

TN — Sat, 09 May 2009 09:33:06 -0000

You should give http://lnk.by/ a try.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

Louis Gray — Fri, 08 May 2009 20:22:08 -0000

I wouldn't say "worried" is the right word. What I was trying to illustrate with this post is that there are many different factors that could impact uptime (or lack thereof). FriendFeed being down today is timely. The discussion of URL shorteners is timely. The conversation of FF.im happened just this week, so it's a good example. Yes, you are right in terms of addressing cloud, hosted services, and survivability. We need to have contingency plans in place for our data, wherever it is.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

guruvan — Fri, 08 May 2009 20:21:56 -0000

I've seen facebook go down in the last couple of weeks, for on the order of about 30minutes or so.

There are good reasons to have accounts on several services if you rely on social media. I have accounts in many many places.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

guruvan — Fri, 08 May 2009 20:19:19 -0000

there are many things that are beyond even the service's control. Carrier backbone BGP failure is one of them. If 2 or three major carriers have major bgp failures this can bring huge chunks of the internet as a whole down.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

awilensky — Fri, 08 May 2009 20:18:33 -0000

You are worried about Friend feed and URL shorteners? We have an entirely new regime of hosting services (cloud, SAAS, PAAS), that is in the process of trying to convince the fat part of the mid market that they can eschew commodity servers in the closet down the hallway, and roll out capital line of business onto elastic cloud services. Now, this will be a conundrum. Until the industry can make an convincing argument (other than keeping completely duplicate systems and comm links), the SME is going to be a tough sell. These are mission critical use cases - POS system, a mid range distributed rapid replenishment system, etc.

Many of the venture-backed cloud start ups have not addressed the survivability issues, and as a corollary, the insure-ability issue. There is not one specialty business underwriter that would touch these under capitalized, thinly equipped, and unrated services. The industry is offering minor alternatives for otherwise existing managed services where the actual, functional differences are virtually indiscernible. Managed services for legacy AS400 applications are far better cases for outsourced remote services.

In order for the industry to thrive, it must address the doubts of the SME, until the outages can be indemnified, there will be no major buy in from the SME. What you can't insure, you can't rely on; someone has to underwrite and price risk - even if the answer is SELF insurance.

FriendFeed is down? Please. Cry me a river.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

tweetfeeds — Fri, 08 May 2009 20:12:48 -0000

the basic architecture of the internet supports always-on scenarios, lots of applications do not, that's the problem.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

guruvan — Fri, 08 May 2009 20:02:00 -0000

right there with ya on this one Jesse. I'm using adjix from now on for my shortening needs ;-) (except of course autosortens from friendfeed.

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

John E. Bredehoft (Empoprises) — Fri, 08 May 2009 20:00:12 -0000

I wonder if we're expecting too little of consumer social media applications.

On the one hand, you can argue that we get what we pay for. Certainly if Twitter or FriendFeed had users that were paying US$50/month to use the service, they would have the obligation to offer a very high level of uptime, and would take care to design their systems to do so. Since both services are in a pre-monetization mode, one can argue that it's not wise to spend money for that kind of resilience at this time.

On the other hand, you can argue that they're damaging their future potential. Those who knew of Twitter during a good chunk of 2008 identified it with the fail whale, and now that there are many more users on Twitter, a fail whale can do a lot more damage to Twitter's image. FriendFeed is still trying to break into the public's eye, and while they've been relatively free of downtime, additional downtime episodes may bring FriendFeed unwanted attention.

I'm one of the extremely small minority of people who use these two services (base on the comments, it appears that most of TechCrunch's readers are on Facebook, or perhaps the WELL), and the only thing that I've determined is that I have to add more of my FriendFeed/Twitter contacts to Facebook. Then again, that could go down too...

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

Jesse Stay — Fri, 08 May 2009 19:43:13 -0000

If they were doing static HTML pages with meta redirects like Adjix does, and hosting on Amazon S3 ff.im and tr.im would still be working. :-) I freakin' love that method of URL shortening - if I ever do a service like that I'm totally using that method. Of course, that relies on Amazon S3 being up all the time. Amazon *never* goes down, right?

Re: http://blog.louisgray.com/2009/05/every-piece-of-infrastructure-carries.html

jeffisageek — Fri, 08 May 2009 19:40:05 -0000

ok louis what did you unplug? :)