Feeds

back to article Google's SPDY blamed for slowing HTTP 2.0 development

The HTTP 2.0 working group appears to be in crisis, with work on integrating Google's SPDY HTTP-boosting protocol blamed for taking the project off the rails. The accusation comes from prominent FreeBSD developer Poul-Henning Kamp, whose post to the working group's mailing list calls for work on the protocol to be abandoned. …

COMMENTS

This topic is closed for new posts.
Anonymous Coward

LULZ

So much for Godgle.

2
4
Silver badge

I stopped seeing SPDY as an alternative...

...when they openly admitted it performed worse than HTTP on high latency and lossy connections... considering more and more people are using wireless connections which are lossy and high latency, I wonder what's the true motivation behind it.

12
0
Silver badge

Re: I stopped seeing SPDY as an alternative...

Latency on wireless depends a lot on what kind of wireless. 3G and below is high latency, LTE, which is what's in Google's backyard, isn't that bad.

SPDY deserves some credit for persistency, better handling of multiple requests and encryption. A working implementation is always better than the best pipe dream. Without SPDY we'd probably still have little prospect of getting off HTTP 1.1

However, the criticisms should be taken seriously. Google normally plays pretty nicely in such discussions - better than most "industry leaders" in any case. Maybe they'll take the criticism onboard and try and fix SPDY if the WG does decide to drop it.

8
1

Re: I stopped seeing SPDY as an alternative...

I've always suspected that SPDY was just a way for Google to save costs.

Given Google's size, even a fairly small advantage in efficiency (bandwidth, cpu power) for only a fairly small part of it's visitors (ie. Chrome users) would mean a large sums of money can be saved.

Giving it away for free only increased their benefits, since other browser makers also implemented it...

1
1
Anonymous Coward

Re: I stopped seeing SPDY as an alternative...

True motivation is to deliver animated 32 bit color adverts to every web page you visit.

No matter how efficient you make the protocols they'll just cram more garbage down the pipe. As a completely wild notion, why not just block the ads by default and let people opt-in if they want them and free up all that lovely bandwidth.

Case in point, the Reg site has gotten a lot slower since they brought Google ads and analytics into each page.

1
2
Thumb Up

Good

Internet protocols should be designed first and then implemented afterwards. Retrofitting the design to an existing implementation is a sure fire way to get lock in.

4
2
Bronze badge

Re: Good

At least it's still an optional protocol, so it can always just be turned off if it's as bad as they say.

0
0
Roo
Silver badge
Windows

Re: Good

"Internet protocols should be designed first and then implemented afterwards. Retrofitting the design to an existing implementation is a sure fire way to get lock in."

I understand the lock-in concern, but in practice doing it back-asswards has panned out pretty well in many cases - Ethernet, IP, SMTP, NNTP, HTTP, NFS etc... At the end of the day IP protocols are about exchanging info between two widgets, protocols that are inclusive (ie: vendor neutral) will tend to be used more widely and a process of selection will weed out the ones that aren't used as widely...

The biggest barrier to avoiding lock in is, surprise, surprise, patents.

1
0
Gold badge

Re: Good

Historically, internet protocols have tended to follow the "two interoperable implementations" rule before being adopted as standards, which is the reverse of what you've just advocated. Design is nice, but very few people can anticipate every detail of how a network protocol will respond on all kinds of real-world network.

0
0

PHD is unfortunately one of those people that know enough about tech to be dangerous, he will decide that night is better than day because you can better see stars at night and therefore anyone doing anything in daylight unrelated to star watching got it wrong.

Makes it hard to know if SPDY has a problem or it's simply a case of PHD having a problem with something else Google has done.

1
3
WTF?

Is there a good summary somewhere of what the 'numerous hard problems' are?

5
0

WWW cannot be saved

Just search 'spdy issues' and you'll find plenty. A big one is that SPDY *requires* SSL/TLS in order to bypass proxies.

Glancing at the mailing list, I see Poul catching flak for other proposals, like "get rid of cookies", which would be great but break ~99% of websites. Thing is, mandatory SSL/TLS is also unrealistic. If that's in HTTP 2.0, most of the web will stay on 1.1.

And SSL/TLS is a security & management nightmare which should be deprecated ASAP, not enshrined in the next generation of protocols.

But anyway, this is all an exercise in turd polishing. The entire WWW is a rough prototype, long overdue for replacement.

2
3
Silver badge

Re: WWW cannot be saved

I just searched for exactly "spdy issues" and I'm none the wiser. A few routine bugs in people's implementations is all I found.

PS given the whine about encryption, do you work for the NSA or GCHQ?

1
0
Silver badge
WTF?

Re: WWW cannot be saved

And SSL/TLS is a security & management nightmare which should be deprecated ASAP, not enshrined in the next generation of protocols.

The council of Elrond demands an explanation for this bullshit.jpg

0
0

Re: WWW cannot be saved

No, do you? NSA fucking loves SSL.

Dat false sense of security...

0
0
Bronze badge

Dog food

Sounds like someone needs to apply a SPDY algorithm to the working group.

0
1
Anonymous Coward

Not really Google's problem

They moved things forward with SPDY whilst HTTP 2.0 working group were dragging their heels.

0
0
Anonymous Coward

evidence

When you say that the fiasco comment ("Now even the WG chair publically admits that the result is a qualified fiasco and that we will have to replace it with something better "sooner”.") is hard to substantiate, you needed to read further in the mailing list where Mark said:

"This is a gross mischaracterisation of what I said. I appreciate that you have a flair for the dramatic, but kindly refrain from putting words in my mouth."

http://lists.w3.org/Archives/Public/ietf-http-wg/2014AprJun/0820.html

0
0
Silver badge
Paris Hilton

Shouldn't people "fix" TCP first?

I regularly hear Bad Things and Possibilities Of Tuning.

1
0
This topic is closed for new posts.