High Scalability

Agustin | 30th November 2011

DataSift is the subject of the latest post on the High Scalability blog which includes a detailed overview of the platform architecture and the problems involved in meaningfully filtering unstructured data from the Twitter API in real time.

‘You have to be able to reliably consume it, normalize it, merge it with other data, apply functions on it, store it, query it, distribute it, and oh yah, monetize it. Most of that in realish-time. And if you are trying to create a platform for allowing the entire Internet do to the same thing to the firehose, the challenge is exponentially harder.’

‘DataSift is in the exciting position of creating just such a firehose eating, data chomping machine. You see, DataSift has bought multi-year re-syndication rights from Twitter, which grants them access to the full Twitter firehose with the ability resell subsets of it to other parties, which could be anyone, but the primary target is of course businesses.’

‘DataSift's real innovation is in creating an Internet scale filtering system that can quickly evaluate very large filters (think Lady Gaga follower size) combined with the virtuous economics of virtualization, where the more customers you have the more money you make because they are sharing resources.’

Architecture In A Picture

‘DataSift has created an awesome picture of their overall architecture. Here's a small version, for the full sized version please go here.’

‘The diagram has two halves: everything on the left is data processing and everything on the right is data delivery. 40+ services run in the system, these include: license service, monitoring service, limit service, etc.’

‘The system as a whole has a number of different scaling challenges that must be solved nearly simultaneously: handling the firehose, low latency natural language processing and entity extraction on tweets, low latency in-line augmentation of tweets, low latency handling very large individual filters, low latency evaluation of a large number of complex filters from a large population of customers, link resolution and caching, keeping a history of the firehose by persisting the 1TB of data it sends each day, allowing analytics to be run on the history of the firehose, real-time billing, real-time authentication and authorization, a dashboard to let customers know the status of their streams, streaming filter results to 1000s of clients, monitoring every machine every filter and every subsystem, load and performance testing, handling high network traffic, and messaging between services in a low-latency fault tolerant manner.’

For more detail on DataSift’s architecture read the full article here.

Previous post: Introduction Streaming API

Next post: Standard and Poor's Downgrades US Banks