Today is an important milestone for our company. We have rebranded the company to reflect that we offer much more than an in-memory database. We also have an expanded vision to share. SingleStore, formerly MemSQL, provides one platform to actualize all of your enterprise’s data.
“What’s in a name?”
The MemSQL name has stood for unrivaled speed, scale, and simplicity for operational data since its founding in 2011. It is a well-recognized name among the data architect and performance engineer experts. Our Co-Founder and Chief Strategy Officer, Nikita and Adam, our CTO, founded the company on the vision of building a massively scalable, elastic transactional system. As our product has expanded to fulfill the growing developer community’s needs, the name no longer reflected the breadth and depth of our current capabilities and product vision. To our faithful communities of users, contributors, customers and advocates, we will “retain that dear perfection” for which you’ve known and loved MemSQL while broadening our capabilities.
The Early Years
The initial version of the product was designed to meet the low-latency requirements of real-time analytics workloads and leveraged the newly affordable large memory hardware available. At the time, the cost of RAM was going down dramatically which made it cost-effective for the first time to build a totally in-memory transactional database where the entire dataset fit in memory. The advantage was blazing speed for concurrent reads and writes, but without the need to manage disk I/O, it also opened the door to using a more efficient indexing approach, namely the skip-list index. Even now, almost a decade later, most databases leverage a less efficient index approach of a bygone disk-based era, the Btree. And so, with an initial version which provided an in-memory transactional rowstore, the company was named “MemSQL” with “mem” signifying in-memory and “SQL” making it clear that you could indeed achieve speed, scale, and SQL without giving up on the expressive power and advantages of relational algebra executing for your application using a simple, widely-understood declarative statement.
NewSQL Pioneers and ACID Guarantees
Within a couple of years, MemSQL became a leader in a new area called “NewSQL” delivering the scalability, speed and flexibility promised by non-relational systems while retaining the support for SQL queries and ACID (atomicity, consistency, isolation and durability) guarantees. As a scalable distributed SQL database, MemSQL was among the first systems to provide these NewSQL capabilities along with lock-free data structures, code generation, MVCC, replication, and k-safety. The next several releases added an in-memory compressed columnstore backed by disk, JSON as a native datatype, times-series and geospatial types and functions, SIMD vectorization, a resource governor, and so many more than we can list here. This brought us into a category Gartner called hybrid transactional analytical processing, HTAP. With the addition of data science and machine learning model integration several years ago, we expanded into an area we called Operational Analytics.
Fast forward to today
The product evolved beyond in-memory several years ago to use a sophisticated tiered storage approach leveraging modern cloud and hardware innovations giving customers 10x the performance at ⅓ the cost of database incumbents like Oracle. Our new name signifies the company’s goals of helping businesses adapt more quickly, embrace diverse data and accelerate digital innovation by operationalizing all data through one platform for all the moments that matter.
What’s Changed – We are now SingleStore
The renaming of the company to SingleStore also brings new product and service names.
The new name is not so new.
Our community of customers and developers will recognize “SingleStore”. It has been the name of the multi-year initiative to move from our dual table type approach in a single database to a solution which both OLTP and OLAP workloads efficiently and with high-speed performance using a single table type. That journey began with the 7.0 release, continues, and will be mature with the upcoming 7.5 release. This capability of a single table type and the initiative is now known as “universal storage” indicating our intent to expand that diversity of data types, data models, built-in functions and data access patterns beyond the current multi-model set we currently support.
Expanded Vision Takes Hold
With today’s rebranding announcement also comes a preview of the expanded vision for the product. Our Chief Product Officer, Jordan Tigani, announced today our intent to provide access to data located anywhere, even beyond SingleStore. The new capability will provide a global namespace for data located across a multi-cloud landscape allowing SingleStore to provide an API from which to operationalize your data no matter where it lives. SingleStore databases will be accessible from anywhere you have a SingleStore compute cluster, while honoring access permissions and sovereignty restrictions. You will be able to join data in AWS against data in GCP and on-premises, for instance. We will manage replication, consistency, and security according to policies. This global federated access to all your data will provide greater flexibility and will further simplify cloud data management for organizations. It’s a game changer.
We’re very excited to share this news with you today during our (R)Evolution 2020 event hosted by our CEO, Raj Verma, and CPO, Jordan Tigani, in collaboration with our customers and partners. Thank you to all of our customers, partners, and employees for helping to build this company and joining us as we continue towards our expanded mission, now as SingleStore, to operationalize all data through one platform for all the moments that matter.