Seems logical and expected. I was wondering if MongoDB would have to pay some licencing fee to WT for using their technology in document level locking and compression. Works well for us Mongo users :)
WiredTiger has the same techie principals as SleepyKat – get the wordplay?! – which was Mike Olson’s company before Cloudera. When asked, Mike spoke of those techies in remarkably glowing terms.
I wouldn’t be shocked if WiredTiger wound up playing the role for MongoDB that InnoDB played for MySQL. What I mean is that there were a lot of use cases for which the MySQL/MyISAM combination was insufficiently serious, but InnoDB turned MySQL into a respectable DBMS.
I think the comparison to InnoDB is apt; many of the same shortcomings/features are there, as well.
MyISAM/mongodb both have table/collection locking, no transactions, make no ACID guarantees. WiredTiger/InnoDB by contrast introduce row/document level locking, transactions, and ACID.
Incidentally, this is why we moved to TokuMX over vanilla Mongo; it provides similar benefits, and is production-ready today. The disk usage is massively improved, document locking means that I practically never have to worry about lock contention (and if I do, it's usually a sign that I have something messed up in my app, rather than something wrong in my DB), I get guarantees about durability, they've even done some heavy work to fix Mongo's election protocol which fixes a number of outstanding issues in failover scenarios. With those features in place, it's actually a decently respectable DB; if WiredTiger can deliver a similar level of improvement, it'll do very good things for Mongo as a DB and a company.
Good for MongoDB. They really need some serious tech people and it seems the acquired team have got the experience, expertise and reputation to lead the technology development.
One thing though: there is one undisputed candidate for the title "the most widely-used embedded data management software in the world", and that is sqlite. BDB is indeed widely used, but sqlite would be deployed in at least one order of magnitude more computers.
I wouldn't be so sure about that. I believe the core database from which BDB evolved (and which was also written by Keith Bostic) is part of some pretty widely-used Unix software (SMTP servers, or DNS, or something like that).
Isn't bdb a standard component on vxworks? Android isn't the most popular embedded thing. I'm just wondering what the actual numbers are on this stuff.
This is fantastic news for MongoDB 2.8. 2.8 should be one of MongoDB's biggest feature releases considering that they have WiredTiger and document-level locking.
My only problem I had with MongoDB in the past was scaling without document-level locking. For every write, my program was waiting for a write receipt, instead of the default fire and forget. I was noticing latency when the whole database would get locked. I'm very excited for document-level locking.
I also have similar skepticism towards MongoDB. I hear more and more horror stories of a team that realy thumped the MongoDB, only to hear they've ditched it completely in favor of PostgreSQL.
Personally speaking, I'm surprised how good PostgreSQL is and I realize that a lot of the reason we use a certain DB is because of marketing. Had I know about PostgreSQL before MySQL I would've never even made the switch.
I'm one of those - after almost three years with MongoDB, I left it behind when I formed a new start-up using PostgreSQL.
I have zero DB issues now, and all the MongoDB experience really worked for my employees as they now work full-time on it at a dedicated MongoDB provider, so win-win.
Great acquisition by MongoDB. Addressing the shortcomings and fixing them like this is really smart. It's great that MongoDB spent the last year and a half improving the codebase and I expect to see great features in the coming years. The story has just begun with those guys I think and as they get more and more adoption in the enterprise arena (as the DB improves) it will become a standard DB across all lines of business.
I know they aren't the most popular thing around here but they are insanely popular and it's great to see the product improving.
19 comments
[ 2.7 ms ] story [ 57.2 ms ] threadWiredTiger has the same techie principals as SleepyKat – get the wordplay?! – which was Mike Olson’s company before Cloudera. When asked, Mike spoke of those techies in remarkably glowing terms.
I wouldn’t be shocked if WiredTiger wound up playing the role for MongoDB that InnoDB played for MySQL. What I mean is that there were a lot of use cases for which the MySQL/MyISAM combination was insufficiently serious, but InnoDB turned MySQL into a respectable DBMS.
MyISAM/mongodb both have table/collection locking, no transactions, make no ACID guarantees. WiredTiger/InnoDB by contrast introduce row/document level locking, transactions, and ACID.
Incidentally, this is why we moved to TokuMX over vanilla Mongo; it provides similar benefits, and is production-ready today. The disk usage is massively improved, document locking means that I practically never have to worry about lock contention (and if I do, it's usually a sign that I have something messed up in my app, rather than something wrong in my DB), I get guarantees about durability, they've even done some heavy work to fix Mongo's election protocol which fixes a number of outstanding issues in failover scenarios. With those features in place, it's actually a decently respectable DB; if WiredTiger can deliver a similar level of improvement, it'll do very good things for Mongo as a DB and a company.
One thing though: there is one undisputed candidate for the title "the most widely-used embedded data management software in the world", and that is sqlite. BDB is indeed widely used, but sqlite would be deployed in at least one order of magnitude more computers.
BDB are used a lot on the server side, but you can't beat the consumer devices.
My only problem I had with MongoDB in the past was scaling without document-level locking. For every write, my program was waiting for a write receipt, instead of the default fire and forget. I was noticing latency when the whole database would get locked. I'm very excited for document-level locking.
Personally speaking, I'm surprised how good PostgreSQL is and I realize that a lot of the reason we use a certain DB is because of marketing. Had I know about PostgreSQL before MySQL I would've never even made the switch.
I have zero DB issues now, and all the MongoDB experience really worked for my employees as they now work full-time on it at a dedicated MongoDB provider, so win-win.
I know they aren't the most popular thing around here but they are insanely popular and it's great to see the product improving.