The Anatomy Of Search Technology: Blekko’s NoSQL Database

[krishna2]

This is the same technology that is used by the webgrepper tool [http://blekko.com/webgrep] (a grep for the web pages' sources).

Disclaimer: I work at blekko and I developed the webgrepper.

As a side note, we have used this for various other purposes - some fun ones being, store a big music collection (to extract meta data via mapjob), citizenship test q&a (to pick random questions), the 'joke of the day' (of course, this is our "hello world" example internally to new employees) ..etc.

[DennisP]

It looks like it'd be useful for a lot of stuff besides search engines. Is there any place to get more details?

[JohnGolt]

Hey, great stuff! How do you implement high availability?

[greglindahl]

There are 2 parts to our high availability.

First is on our frontend side. We have 2 nginx servers (using linux HA and vips) which send traffic out to the nodes of the cluster which are up, retrying to a different node in the case of failure or a slow reply.

Deeper in the system, there are 3 copies of every piece of data.

Both of these are fairly normal mechanisms; the 3-copy thing is used at Google and by Hadoop and friends.

[jsrfded]

Within the datastore, there are 3 copies of each piece of data. When a get() request is made, it goes out to the "closest" copy; if an answer isn't heard from by some threshold, a 2nd request is made to one of the other replicas. Whoever gets the data back first wins.

[kveykva]

Isn't DynamoDB on SSDs? AWS alone might not be cost effective but, how does it look with the other services also offered?

[wumpus]

Since our SSDs hold a cache of what's on disk, it's most effective to have the SSDs and the disks are in the same server. It's hard to assemble separate pieces into a tightly-coupled system.

[Juha]

Very good article in deed. I have always been amazed of how the search engines can query so huge data sets so quickly. This brings some light to it.

It would had been nice to see some examples of the query language they use, it if is comparable to other NoSql databases.

[wumpus]

Like many NoSQL databases, we don't really have a query language. Mostly you read a few columns from a particular row. That could contain a combinator, such as at TopN containing 1,000 inlinks to a webpage... something that you would ordinarily fetch using a range query that returned 1,000 rows from a table.

[gruseom]

Can you say more about the choice of swarm algorithms instead of Paxos?

[jsrfded]

Paxos would be good for electing a master, but we wanted to avoid having any masters in the architecture. There are also scenarios where paxos can be slow or fail to reach a consensus. We wanted high availability from each node in the cluster regardless of whether 2/3 of the rest of the cluster were down or unreachable; both parts of a partioned cluster should also be able to continue to function as best they could.

Individual nodes can often make "personal" decisions about what to do in subobtimal situations. If you can answer an incoming request, even with partial or out-of-date data, do so; it's better than not replying. For the repair agent, each node can see its own view of "holes" in the 3-level replication, and offer to make copies of <3 buckets to bring back up to three copies.

[lsuejung]

Awesome insights into what goes into building a search engine. Very impressive indeed.

[tinyjoe]

does blekko DB partition data base on primary key or it just query all nodes everytime like elasticsearch?

[jsrfded]

Partitions are based on a hash of the primary key. The number of buckets in the system has to be a power of 2. But we can split buckets to increase the number, or even have some buckets that have split and some that haven't yet. Each bucket is stored on 3 separate servers (and the assignment makes sure the three servers are on separate racks).