Forgot your password?

Comment: More likely case (Score 1) 125

by rekoil (#46071387) Attached to: How Google Broke Itself and Fixed Itself, Automatically

What's more likely - I've run into exactly this scenario before, in fact - is that the configuration generation system regenerates configs on a regular schedule, and at one point encountered a failure or spurious bug that caused it to push an invalid config. On the next run - right as the SREs started poking around - the generator ran again, the bug wasn't encountered, and it generated and pushed a correct config, clearing the error and allowing apps to recover.

Comment: Re:Silly priorities (Score 3, Informative) 274

Disclaimer: Another Twitter engineer here. What my apparently former colleague said, plus X.

Also: Don't be afraid to add caching layers when you see your web server or DBs start to run hot. Putting a memcached instance in place in "front of" your database layer is much easier than sharding the database layers to relieve load - eventually you'll have to do both, but you'll definitely want the memcache layer first. Same with web caches/proxies - putting varnish or squid in front will take some pressure off before you need to implement load balancers.

Comment: Re:Not so fast...YET (Score 1) 135

by rekoil (#40356211) Attached to: SPDY Not As Speedy As Hyped?

In fact, if SPDY support was ubiquitous tommorrow, I would be surprised to see SPDY+TLS used for third party ad serving for this very reason.

Good news here: Google's DoubleClick and AdSense ads are served over SPDY today. In fact, I'm not aware of any Google properties that don't use SPDY, since they're all routed through the same GFE (Google FrontEnd) proxy farms.

Comment: Re:The problem with the test ... (Score 5, Informative) 135

by rekoil (#40352749) Attached to: SPDY Not As Speedy As Hyped?

SPDY as implemented requires SSL, since the protocol capability is negotiated by a TLS extension on port 443. There's no spec for negotiating SPDY on a standard HTTP port - it would only work if the capability was assumed on both sides before the connection (for example, URLs that start with spdy:// instead of http:/// which connects to a different TCP port on the server).

Comment: Re:Single domain? (Score 1) 135

by rekoil (#40352719) Attached to: SPDY Not As Speedy As Hyped?

That only works if all of those hostnames resolve to the same IP addresses. The main optimization in SPDY is the elimination of the need to make multiple TCP connections simultaneously, but all of those resources must live on the same server. If the resources have different hostnames, you might be able to detect hostnames that point to the same IP and then interleave those, but I don't know if the current implementations do that yet.

Most CDNs, however, return different IPs for nearly every query, and web developers use multiple hostnames pointing to the same resources to get non-SPDY multiplexing today. This sounds like an optimization that's easy to accomplish dynamically, though (if request is SPDY, don't spread the resources across different hostnames).

Comment: Re:What is the big deal with SPDY? (Score 5, Informative) 135

by rekoil (#40352679) Attached to: SPDY Not As Speedy As Hyped?

1. HTTP Pipeline support proved very difficult to implement reliably; so much so that Opera was the only major browser to turn it on. It can be enabled in Chrome and Firefox but expect glitches. By all accounts SPDY's framing structure is far easier to reliably implement.

2. WIth SPDY, it's not just the content that's compressed but the HTTP headers themselves. When you look at the size of a lot of URLs and cookies that get passed back and forth, that's not a insignificant amount of data. And since it's text, it compresses quite well.

3. SSL is required for SPDY because the capability is negotiated in a TLS extension. Many people would argue that if this gets more sites to use SSL by default, that's a Good Thing.

4. If you're running SPDY, the practice of "spreading" site content across multiple hostnames, which improves performance with normal HTTP sites, actually works against you, since the browser still has to open a new TCP connection for each hostname. This is an implementation issue more than an issue with the protocol itself; I expect web developers to adjust their sites accordingly once client adoption rates increase.

5. The biggest gains you can get from SPDY, which few have implemented, is the server push and hint capability; this allows the server to send an inline resource to a browser before the client knows it needs it (i.e. before HTML or CSS is processed by the browser).

But as someone else as pointed out, the author's test isn't really valid, as he didn't test directly to sites that support SPDY natively, he went through a proxy.

The website I work for is supporting SPDY, and the gains we've seen are pretty close to the ~20-25% benchmarks reported by others. As many have pointed out, this author's methodology is way broken. I'd recommend testing to sites that are known to support SPDY (the best-known are Google and Twitter), with the capability enabled and then disabled (You can set this in Firefox's about:config, Chrome requires a command line lauch with --use-spdy=false in order to do this, though).

Never buy from a rich salesman. -- Goldenstern