Late in the week we learnt that Google has acquired privately held Talaria, based in Palo Alto, a company with less than 20 employees. [Talaria, is Greek for sandal, a symbol for the Greek god Hermes - and with some fantasy you can see a sandal in the logo below].
![]()
![]()
Talaria meets... Google
Not much was known about Talaria before this acquisition, the company had a one page only website - which even disappeared after the news of the acquisition was out on Friday. Thanks to Google's Internet caching though - it was easy to see what Talaria claimed to be up to:
The product idea is to improve website performance, and the trick is to move code from more popular, high level programming language via a JIT compiler into machine code. Talaria claims to have done this successfully for PHP code being generated by Wordpress, as well as for the content management engine Drupal.
Talaria was in private beta and soliciting sign ups from their (now defunct) website. They built a web server that seamlessly integrates into an existing web architecture by replacing the scripting engine. With support of FastCGI, there is standards based support to speak to a variety of web servers.
Talaria has also been smart about productizing intermediary products: Naturally, when compiling code to byte language, you need to analyse and parse the higher level code. That allows you to profile code and pin point to non working (it simply won't compile) or badly performing code (that's the one you may want to move first to byte code). So Talaria productized this functionality into a Profiler. Smart.
The Talaria business case
If you are able to replace slower, often only interpreted code from popular programming languages such as Phyton and Ruby - you gain dramatic efficiencies, often even better than the 10x Talaria claims. That allows you to either push your functionality further on existing hardware - or reduce the hardware needed to run the same functionality. Pick your business case - it works.
Talaria will become part of the Google Cloud platform, where it will create the same benefit - with the nice differentiation for the Google Cloud sales team, that they will be able to tell customers and prospects that they can keep coding in e.g. Ruby, but on Google Cloud platform they will be able to run them faster than anywhere else. This creates much desired stickiness and value for Google Cloud platform - both for their customers and Google themselves.
And with Google running more than a million severs [side note - that's approx 20 per employee (!)] - makes a pretty good business case on the side.
That there is a business case transforming slow running e.g. PHP code to a closer to machine language, has been proving by Facebook's (now open sourced) HipHop PHP to C++ converter) - a little more than three years ago. It may have inspired the Talaria founders, even to the point that they built on HipHop. Only speculation.
Why does Google buy a company in private beta
Talaria must have been up to something very good... And its founders, Austin Robison and Solomon Boulos (now LinkedIn job title- person who builds things at Google) must be two pretty smart engineers, which works always well with the Google culture (see Boulos impressive publication list at Stanford here).
For Talaria the road to success was pretty steep - in order to embed another web server, you need to trust that it generates pretty good code - and that takes some time and hand holding before anyone makes the switch. And then there are a lot of languages to parse, many web servers to embed etc. Not sure Talaria had the funding for that. The support for NGINX also points to some potential russian adventures in funding [Speculation]. Now things get easier - as the Talaria web server only needs to work for the Google Cloud platform - and creates instant value, starting with PHP.
MyPOV:
When Google acquires a small and unproven company, they must be acqhiring smart people. There are significant benefits for Google both internally and externally, by building out the Talaria web server for Google Cloud. Look at Amazon's AWS to either finish internal work or acquire another JIT vendor soon.
And lastly: Hopefully a nice exit for two smart University of Utah alums, may they and their product idea thrive well inside the Googleplex. If they succeed the cloud will be the winner.