Skip to main content

How To Expose the Guts of Twitter (A post about Starling)

Twitter does a lot of queuing. I mean, a lot. We know other people have a need for some good queuing, so much that Amazon even released Amazon Queue Service, not so long ago. There has never really been a common queue server, and maybe that is because its so simple that no one has really had the need to push one hard into the public eye. At least, as public as our eyes are.

Enter Starling, the internal queue system of Twitter, recently released to the public. Written in Ruby, and I don't even mind! Pointed there by my ever-pointing buddy, David Novakovic, Starling does nothing absolutely remarkable, but someone has to get the light. What is interesting is their choices. Starling uses the MemCached protocol, so your clients are probably already prepared to use it, they just need to treat the queues a little different from the mappings. The typical MemCached get-operation now removes the item from the queue. The keys function is identifiers for the queues. I don't think it could have been simpler. I'm planning to look at setting up Starling for testing on my linux servers and my Macbook, and to try and find something interesting in the way of using it. I have some plans I could utilize it in, and maybe bring it to the office later.

Now that Starling has some attention and gives us something of a standard for queue protocols (I love reusing protocols!), if anyone has different needs or just wants to scratch an itch in their language of choice, lets make the smart move and take the same protocol route. Queues may be a small thing, but its the same things we really need to agree on more. Anyone up for a MemCached-protocol to Amazon Queue Service bridge?

Comments

Jamieson said…
I wouldn't say that distributed, "reliable", low-latency queuing is exactly simple, although the concept is...

But also check out XMPP (which does wonderful queuing and presence) and AMQP... XMPP is well represented by ejabberd (Erlang), jabberd, etc.. AMQP is more-or-less represented by Apache ActiveMQ, RabbitMQ, etc. AMQP is more traditional, whereas XMPP is more SQS style.

https://stpeter.im/?p=2099

Popular posts from this blog

Interrupting Coders Isn’t So Bad

Here’s a hot take: disrupting coders isn’t all that bad.

Some disruptions are certainly bad but they usually aren’t. The coder community has overblown the impact. A disruption can be a good thing. How harmful disruption might be a symptom of other problems.

There are different kinds of disruptions. They are caused by other coders on your team, managers and other non-coders, or meetings throughout the day.

The easiest example to debunk is a question from a fellow developer. Imagine someone walks over to your desk or they ping you on Slack, because they have “one quick question.” Do you get annoyed at the interruption when you were in the middle of something important? You help out your teammate quickly and get back to work, trying to pick up where you left off. That’s a kind of interruption we complain about frequently, but I’m not convinced this is all that bad.

You are being disrupted but your team, of which you are only one member of the whole unit, is working smoothly. You unstuck …

Announcing Feet, a Python Runner

I've been working on a problem that's bugged me for about as long as I've used Python and I want to announce my stab at a solution, finally!

I've been working on the problem of "How do i get this little thing I made to my friend so they can try it out?" Python is great. Python is especially a great language to get started in, when you
don't know a lot about software development, and probably don't even know a lot about computers in general.

Yes, Python has a lot of options for tackling some of these distribution problems for games and apps. Py2EXE was an early option, PyInstaller is very popular now, and PyOxide is an interesting recent entry. These can be great options, but they didn't fit the kind of use case and experience that made sense to me. I'd never really been about to put my finger on it, until earlier this year:

Python needs LÖVE.

LÖVE, also known as "Love 2D", is a game engine that makes it super easy to build small Lua…

CARDIAC: The Cardboard Computer

I am just so excited about this.


CARDIAC. The Cardboard Computer. How cool is that? This piece of history is amazing and better than that: it is extremely accessible. This fantastic design was built in 1969 by David Hagelbarger at Bell Labs to explain what computers were to those who would otherwise have no exposure to them. Miraculously, the CARDIAC (CARDboard Interactive Aid to Computation) was able to actually function as a slow and rudimentary computer. 
One of the most fascinating aspects of this gem is that at the time of its publication the scope it was able to demonstrate was actually useful in explaining what a computer was. Could you imagine trying to explain computers today with anything close to the CARDIAC?

It had 100 memory locations and only ten instructions. The memory held signed 3-digit numbers (-999 through 999) and instructions could be encoded such that the first digit was the instruction and the second two digits were the address of memory to operate on. The only re…