Nielsen: Hurdling Operational Issues
Henry Snow – VP, Infrastructure, Nielsen Marketing Cloud, Nielsen
My name is Henry Snow. I’m VP of infrastructure for the Nielsen Marketing Cloud. We are a subgroup inside of Nielsen that focuses on connecting the dots between publishers and marketers by enhancing digital data. Aerospike allows us to have read and write transactions on a NoSQL database in sub-millisecond time. This is very important for us as members of the Ad Tech community because all of our transactions need to be as short as possible. We try to complete all of our transactions in less than 250 milliseconds. This goes reading and writing from Aerospike, network time leading to our data centers, as well as any sort of analytics or processing that we do on the data. Aerospike is responsible for storing our user history, either cookie or mobile ID. We’re able to use the history that’s stored inside of Aerospike to make decisions on the new data that’s being ingested into our system.
And then we’re able to store that any sort of decision that’s made for the next time that we see this device. Aerospike allows us to match these IDs to a object that fits most user history for the devices that are calling our system. And we’re able to retrieve process that data in milliseconds. We needed to fill out our offering inside the EU and provide low latency failover capabilities. Aerospike identifies write hot keys very well because that’s a transactional issue where you’re trying to write too many times to the same object over and over and over again. One thing I love about Aerospike is that it’s very capable of taking advantage of any hardware you throw at it. This might be your network devices, this might be your SSDs or your NVMe storage, it might be your RAM.
Together with Aerospike support, we were able to benchmark and identify a certain NVMe devices that will allow us to perform orders of magnitude better than traditional SSDs. I have trouble deciding what I liked most about Aerospike. There’s a bunch of features that really stand out as A-plus features. Cross Datacenter Replication is vital to our organization. We need to have redundant data centers. We need our user objects to be available in multiple facilities. Aerospike does this with ease. It’s able to take advantage of a 20 gigabit link without a problem and ship data continuously. I like that Aerospike integrates into the Linux ecosystem with ease. Servers are able to join clusters using traditional service or a systems ETL commands. Troubleshooting it is super easy using the available ASADM or AS-info tools. Monitoring statistics are readily available.
Aerospike has been a great partner over the years. We’ve reached out to support and the architect teams many times. They respond almost immediately and they’re able to solve our problems within days. Aerospike has rolled out a couple of features ahead of us even realizing that we will use them and need them. One example is the device read cache. This allows us to alleviate some read hot keys, by using the built-in Linux cache subsystem, and improve read performance drastically. The ability to replicate data across region is something that Aerospike provides that very (few) other NoSQL databases do with ease. Without Aerospike, we’d be looking for a new NoSQL data store that performs as well as Aerospike does, and I haven’t seen one out there.er than Aerospike that can currently meet that objective.