Facebook stores data in one data center and processes 500 TB of data per day
Vice President for Infrastructure of the largest social network in the world, which is experiencing hard days on the stock exchange, Jay Parikh told reporters about some statistical indicators of his company, which allows us to assess the scale of the data and the number of users actively working with Facebook.
First of all, it turned out that unlike many companies that build a distributed network of data centers to facilitate the delivery of content to users from different regions, Facebook stores all the data in one “cluster” of more than 100 Pb, which is more than any of the clusters of company data centers comparable in scale to Facebook. This allows engineers to quickly access the network and monitor its status.
Many functions of Facebook require prompt scanning of user data - for example, it is extremely urgent to put friends' status updates and other news in a “chronicle” or offer possible “friends” - for this, the network infrastructure processes 105 Tb of available content every half hour.
In general, the main statistics of the user base are as follows:
It goes without saying that Perikh claims that Facebook engineers make every effort to get the network working as fast as possible and improve the user experience - the data is processed “almost in real time”. “We cannot allow the photo to be uploaded for a whole week,” he says.
[ Source ]
First of all, it turned out that unlike many companies that build a distributed network of data centers to facilitate the delivery of content to users from different regions, Facebook stores all the data in one “cluster” of more than 100 Pb, which is more than any of the clusters of company data centers comparable in scale to Facebook. This allows engineers to quickly access the network and monitor its status.
Many functions of Facebook require prompt scanning of user data - for example, it is extremely urgent to put friends' status updates and other news in a “chronicle” or offer possible “friends” - for this, the network infrastructure processes 105 Tb of available content every half hour.
In general, the main statistics of the user base are as follows:
- Every day 2.5 billion likes are recorded both in the network itself and in social plug-ins on third-party sites
- Also, users upload 300 Tb of photos per day, which until recently was not so easy to delete.
- It generated 70,000 inquiries from people and automated systems ( 70,000 queries Executed by people and automated Systems )
- The amount of data on the network every day increases by 500 Tb
It goes without saying that Perikh claims that Facebook engineers make every effort to get the network working as fast as possible and improve the user experience - the data is processed “almost in real time”. “We cannot allow the photo to be uploaded for a whole week,” he says.
[ Source ]