K2View helps customers democratize data access, presenting a real-time 360-view of any business entity, such as customer, product, work order, supplier, or credit card. The company’s Data Product Platform manages these “data products” as secure micro-databases, populated in real-time by multiple siloed systems across an enterprise and instantly accessible to authorized data consumers. With the Data Product Platform, customers have a trusted, holistic, and real-time view of any entity important to the business.
We recently spoke with K2View CTO Yuval Perlov about the company’s beginnings, its future, and why it decided to work with DataStax as an early adopter for DataStax Astra DB.
1. Tell us about K2View and what sets it apart.
When we looked at the landscape of data migration, data movement, and digital decoupling, we saw that none of the current solutions offered real-time tooling and storage engines. All the big data solutions, data analytics, and big data lakes were a few hours behind at best. For real-time operational use cases, we offer a tailored platform.
For example, we help our customers gain real-time, single views of their customers, and gain operational intelligence by predicting churn, checking credit, and detecting fraud in real time. We also assist with provisioning to test data on demand to accelerate development and assist with data pipelining, privacy, governance, and compliance.
2. Let’s focus on K2View’s beginning. What motivated the company to take on these challenges on such a broad scale?
Our journey started with migration use cases, trying to solve the problem of data migration with no down time. At the time, the primary approach to migrate data was by tables. This means you cannot use the source or target system until you finish writing the last byte of the last table. Our approach was to model the systems into the business entities (such as Customer) they represent and migrate them one by one. This offered reduced down time but also countless other advantages such as better performance, less database locking, faster dev cycles, better data quality and more. Since then we keep finding, along with our customers, new use cases for this approach.
3. Why did you choose to use Apache Cassandra®?
When we started out, everything was on-prem. We serve huge enterprises, so we needed multi-data center support and extremely high availability, and Cassandra was our choice. We built our platform completely on Cassandra as an on-premises solution
4. What issues did you face when moving to the cloud?
When we started shifting to the cloud, we looked at our options, especially on the storage engine. One was to leave Cassandra altogether, go to another engine, or maybe use the native cloud providers' databases.
We looked into Azure Cosmos DB and Amazon Keyspaces, which both have a Cassandra facade. We didn't get the same performance profile we got from Cassandra though, so we would need to change optimizations, and we didn't want to go down that path.
Vendor lock-in is another big obstacle for us. We need platform flexibility. Our promise to customers is that they'll be able to move between clouds, so they don't have to deal with cloud lock-in.
5. Why did you start working with DataStax?
Astra DB from DataStax was very compelling because it's multi-cloud and elastic in its deployment model with a consumption-based structure. So that, in combination with our existing code base that was completely optimized to Cassandra, made Astra DB a natural choice for us. It gives us a lot of flexibility on how we deploy our platform.
Now that we've completed this cloud migration, we're heavily reliant on Astra DB to give us cloud-agnostic service. Using Cassandra as a whole gives us multi-premises flexibility so we can run on-prem, hybrid solutions, and completely on the cloud with any cloud provider where Astra DB is available.
Moving from Cassandra to Astra DB was very easy, and it's almost the same code base. Any other alternative meant rewriting our application and living with difficult cloud lock-in. Cassandra to Astra DB was a great move.
6. Did you consider managing Cassandra yourselves?
We always had the option to manage Cassandra ourselves, but that came with two issues. First, we would be taking on a huge management burden, and second, Cassandra is extremely scalable, but it's not elastic. We would need to provision our services to the highest possible throughput that’s needed. Astra DB effectively gives us elasticity.
The fact that with Astra DB it's real Cassandra behind the scenes and real open source means that we are kind of future proof. We don't have to rely too heavily on custom technology or proprietary technology, and we can decide in the future to move to our managed services, to move to some other company's managed services. Of course, at the moment, we're extremely happy with Astra DB, so we want to stay.
7. What value do you see in Astra DB’s consumption-based pricing?
We’re hearing quite often from our customers that they want to pay for exactly what they use. They're even willing to pay a higher premium during peak hours, knowing that when they are not using the system, it's not costing them. Moreover, it is difficult to estimate hardware capacity needed for new projects, creating unnecessary friction. It can be very hard to estimate project costs.
With consumption-based billing, customers know that they're paying for exactly what they want at any given time. Plus, once they know how to calculate costs, it becomes obvious what's going to be your bill. Customers also don't have to commit to anything beforehand to provision hardware that they might not need.
Astra DB makes it easy for us to offer consumption-based billing because Astra DB uses the same model.
8. What advice would you have for other organizations considering DataStax?
When you set up a production environment, it's okay if it takes a few hours to install and set up. If you need to provision environments all the time for development and QA, you need speed. Astra DB can provision new services and decommission them when not needed. For example, you can provision an environment on the fly as part of a Jenkins Pipeline - a consolidated set of instructions for repeatable building and testing applications. This provisioning is not available if you are running Cassandra on your own.
9. How do you see your future with DataStax and Astra DB?
First, when it comes to cloud provisioning, I think we're going to see much more complex deployment scenarios. We are seeing requirements for hybrid cloud that are completely flexible in their scalability. We are seeing requirements to put clusters up on demand, to do a batch and decommission them completely, but still have them part of an overall sticky cluster. It's all in the name of getting real-time data faster for multiple sources with different synchronization strategies. Customer expectations are just growing, and we need to meet them with very flexible technologies. We see great scalability and performance with Astra DB. And the fact that it has presence in a lot of availability zones throughout our customer cloud landscape is a big plus.
Digital Champions are the builders of data-driven, high-growth businesses. They are the visionaries and driving forces in using real-time data and the cloud to deliver unprecedented value to their organizations. Hear how these Digital Champions like Yuval Perlov are building powerful real-time applications that are making the difference.
