99.96% Uptime is Bogus Marketing

Reliability Update

Twitter has been making great progress in terms of uptime and
reliability. Fail Whale sightings are far less frequent these
days thanks to our efforts but we still have a long journey
ahead. Last month we saw 99.88% uptime and so far this month we
are at 99.96%. Our engineering and operations teams have been
taking a very methodical approach to improving Twitter. We’re
using the word “craftsmanship” to characterize our work here at
the office. Reliability and dependability continue to be top on
or list of key goals.

The above passage is from an email from Biz Stone at Twitter today. After a horrid June, things could only go up at Twitter HQ. Fortunately, it looked like they got serious about the uptime issues they had their and things have been better.

In the meantime, they purchased the super reliable and speedy Summize and branded it with Twitter branding at search.twitter.com.

This could only be a good thing, right?

Well, you’d think. Except the purchase of the super speedy and efficient Summize has only driven the tool into the pond. To be fair, it’s not horrible, but it suffers from the same weaknesses that Twitter does.

That is, it can’t keep up.

As an example, I’ve been following #dnc08 and #rnc08 searches on Summize to watch what people are saying about the political conventions. During the high traffic tweet windows during the evenings of the conventions, Twitter is reliable. That is, they are reliably late. Usually 1-2 hours behind the actual tweet stream.

This is completely unacceptable, and it is complete spin, I guess in the spirit of the conventions, for Biz to tout 99.96% uptime.

Let me be clear, when things are slow and not performing up to standard, you cannot claim 99.96% uptime. Technically you can. Uptime is technically defined by if the web server serves a 404 Page not Found (or Twitter Fail Whale in their case) or a 200 Page found status code.

But from a common sense user experience, this is not uptime. And to claim so is disingenuous.

I appreciate the efforts Twitter has put into improving, but why are we fighting the ability to use the tools during high-demand times. In essence, that makes the tools completely useless.

I look forward to better results, but my skepticism remains in place about Twitter. They do not have the staying power to make it.

I have been on a 3 month hiatus on Twitter blogging. I have refused to blog about it, but there’s another post that has been in the back of my mind for some time. What happens when companies and businesses trying to use Twitter as a marketing and communications tool cannot. What happens when your brand relies on the communication lines and those communications lines dying?

Another day, another post. For today, the spin needs to be exposed.

Published by

Aaron Brazell

Aaron Brazell is a Baltimore, MD-based WordPress developer, a co-founder at WP Engine, WordPress core contributor and author. He wrote the book WordPress Bible and has been publishing on the web since 2000. You can follow him on Twitter, on his personal blog and view his photography at The Aperture Filter.

2 thoughts on “99.96% Uptime is Bogus Marketing”

  1. Most service providers with SLA start at 99.95% for (obviously) paying services. After the SLA is missed, the remedies and credits begin. The flip of this is bonus for outperforming a minimum committed performance metric.Application specific SLA metrics are a huge concern of any provider giving out internal performance guarantees for transaction based systems. It's a more refined way of saying, that just because you are UP doesn't mean your service is USABLE. This is a shift from the availability as a feature and more in terms of functional.Do you have an algorithm to present as the total weighted mix to arrive at a “user experience”?;-)If you do, it's likely patentable. I'd say it is safe to say that the @biz claims are 99.96% accurate.I just get the feeling that their team is tracking a lot of spinning plates right now for the sake of development as priority and infrastructure secondarily. Once they are bought by Google or someone thick with back office the keeping up and other Twitter complaints will shift to the want for features when functional interfaces are taken for granted.

  2. I can't fathom how Twitter could snatch Summize and then throttle it back. I see half-hour delays routinely.And I can't blame it on extra Summize traffic, either – because there's still no link to the Search.twitter function from the main page.How does this happen?I am glad to know that the Twitter team can now watch a movie together. I'll bet Jeff Bezos is proud.

Comments are closed.