Momento emerged from stealth mode with serverless data caching technology designed to accelerate databases.
The vendor, based in Seattle, also said it raised $15 million in seed funding to help build technology and go-to-market efforts.
Momento is led by co-founder and CEO Khawaja Shams, who had previously worked at AWS developing the Amazon DynamoDB cloud database.
At AWS, Shams said he often encountered challenges for users in configuring and managing data caching for DynamoDB and other databases.
The purpose of a cache is to let users rapidly access commonly queried data from a database. Shams left AWS and co-founded Momento in 2021 to build out a serverless data cache platform.
A serverless data cache is a managed service that can help organizations use caching without mnganagi the details of the caching technology.
There are multiple other data caching technologies in the market today, including Redis, Amazon ElastiCache and the open source Memcached technology.
Among the early users of the Momento technology -- formally introduced on Nov. 2 -- is CBS Sports, which is using the serverless data cache for its user registration database as well as its data information tier that provides scores and updates.
Edwin Rivera, principal architect at ViacomCBS/Paramount/CBS Sports, said his team uses multiple databases, including PostgreSQL, MySQL and MongoDB.
The IT team was previously using Memcached and Amazon ElastiCache but sought a more managed service, which is what Momento provides. As it turns out, Momento is faster as well, Rivera said. "We found in our testing that the Momento system was 15% faster than our ElasticCache setup."
Why serverless data caching matters for database operations
In high-data-query-volume environments, like the one Rivera manages for the sports network, caching is an essential service.
"We try to protect our database with caching as much as possible," Rivera said. "We're not built to serve data from disk."
Edwin RiveraPrincipal architect, ViacomCBS/Paramount/CBS Sports
Serving data queries directly from a database is not a best practice for Rivera and CBS for a number of reasons. Database operations are among the most expensive pieces in a data stack. In contrast, operating a data cache is cheaper than scaling up a database, Rivera said.
Running an application that is 90% cached on a cluster of four databases is a lot more cost-effective than having to scale up the cluster, he said. CBS would need about 16 databases in a cluster to handle the same amount of traffic as they do with four databases with data caching.
The only traffic that should actually travel back to the database are data writes, Rivera said. "Generally speaking, our reads are over 90% cached. So that's how we tend to operate. And then our [writes] are the only other things that should make it back the database."
How the Momento serverless data cache works
The vendor uses what is known as a look-aside caching pattern.
With a look-aside cache, when a data query comes into a platform, the system will first look to see if the information is already in the cache. If it's not, the result will be pulled from the database and then also pulled into the cache so subsequent queries can be served faster.
Momento uses the open source Pelikan caching technology that was developed by Twitter as a foundation for its service.
"Pelikan is a hidden gem," Shams said. "It's a rewrite of Memcached built by Twitter."
Shams said his team has worked a lot with the Pelikan project to optimize the technology. On top of Pelikan, Momento has built its own open source protocol for optimizing database caching workflows.
Momento's overall goal for its technology is to be a serverless data caching platform that accelerates databases so they make user interactivity in data-driven applications as fast as possible.
Interactivity in applications is what drives conversion -- whether sales or converting a casual user to a regular user -- in any type of business, Shams said. The ability to deliver data quickly is key to enabling that rapid interactivity. But it can't be done with a database alone.
"Customers are just really eager to reduce their database cost and have the ability to absorb spikes by adding a cache as well," he said.