Matillion Founder Recounts Midrange Roots
February 24, 2021 Alex Woodie
In the mad rush to build data warehouses in the cloud, there’s one group that’s doing surprisingly well: ETL tool providers. Despite the sophistication of today’s AI and analytic endeavors, ETL tools are a critical link in the chain to the cloud. And one of the up-and-coming vendors in this space is Matillion, which last week raised $100 million in venture capital.
Matillion was founded in 2011 to help companies build cloud data warehouses. The Anglo-American firm developed expertise in taking companies’ corporate data from on-premise systems and building data warehouse on emerging cloud systems, such as Amazon Web Services’ Redshift. Around 2014, the company realized there was an unmet need for enterprise-grade cloud-native ETL (extract, transform, and load) software, and so it pivoted to address this need by writing shrink-wrapped software.
Today, Matillion is focused squarely on serving the ELT needs of large enterprise customers moving data into cloud data warehouses. According to Matthew Scullion, the CEO and founder of Matillion, the company’s product success stems to three technical advantages.
First, the software is built for the cloud. Redshift was the company’s primary target, but today its biggest customer base is with Snowflake, where it has 400 customers alone. Databricks, Microsoft Azure Synapse, and Google Cloud’s BigQuery are also familiar end points for its customers’ data.
Secondly, Scullion says, Matillion leverages an ELT architecture, as opposed to an ETL architecture (although the company still refers to itself as an “ETL” provider, proof of that term’s enduring nature). That means that, instead of performing the all-important data transformation step in a secondary server between the source and the target, it uses the processing power of the target server — or, in this case, the massive clusters powering public cloud data warehouses.
Lastly, Matillion embraces a low-code/no-code user experience for its customers. “You can think of it a little bit like Google Docs for data integration,” Scullion says. “That allows people to go faster. In the enterprise setting, importantly, it’s great for maintainability and shareability.”
The enterprise nature of Matillion’s product and its customers base separates it from less capable ETL tools, says Scullion, who has weathered the COVID-19 pandemic near the company’s Manchester, UK, headquarters (it also has a headquarters in Denver, Colorado). Today, large enterprises are the bread and butter of Matillion’s customer base, and represents 70 percent of its revenues, he says.
“Because we focus so concertedly on the enterprise side of the house, we provide really deep functionality, particularly in the on-prem and legacy world,” Scullion says. We have world-class file and relational database management system, a world-class ability to participate in a sophisticated IT setting. If you were a three-year-old company and everything was a software-as-a-service application — we can find those as well. But in the enterprise, with midrange and mainframe systems” Matillion really shines.
Scullion, who spent years as an AS/400 programmer before founding Matillion, says the company makes a point of supporting the Db2 for i database with its Db2 for i connector, which supports the capability to move Db2 for i data into the major cloud warehouses, including Snowflake, Azure Synapse, Google Cloud BigQuery, and AWS Redshift.
To be sure, Matillion isn’t the only ETL tool that supports IBM i and mainframe sources. IBM supports these data sources with its DataStage product. There are also options from Talend and Informatica, which have supported the IBM i server for years with their ETL products, as well as from HelpSystems, which acquired IBM i ETL specialist Coglin Mill several years ago. If real-time data integrations are your need (which is a different, but related, product category) then change data capture (CDC) solutions from Syniti (formerly BackOffice Associates, which bought HiT Software) and Qlik (which acquired Attunity) are available. Precisely (which bought Syncsort and its Vision Solutions offering) bills its Connect offering as a data integration, CDC, and ETL tool.
But it’s good to have options, and for IBM i shops, Matillion definitely provides an option for those seeking ETL tools to get valuable corporate data into cloud data warehouses. It’s been years since Scullion worked on an IBM i server, but he says he still remembers some RPG commands, which puts him in a very exclusive club among CEOs of software companies.
Matillion boasts a collection of more than 100 pre-built connectors, giving customers the ability to pull data out of the Big Iron systems that still dominate corporate data centers, as well as newer sources, like NoSQL databases (Apache Cassandra, Redis, Elasticsearch), relational data stores (Hive, MariaDB, Netezza), SaaS apps (Salesforce, Google Analytics, Marketo), file systems and blob stores (S3, HDFS, GCS, Azure Blob Store), APIs (REST, OData), and social media (Twitter, Facebook, Instagram).
Don’t be surprised if Matillion uses some of the $100 million it raised from its Series D round last week to bolster its suite of connectors. Having pre-built integration points is one way that ETL providers are looking to differentiate themselves in an increasingly hot market for cloud data integration.
“It’s a connector race,” Scullion says. “Everybody is building them. [But] it’s a real pain for everybody involved. It’s a pain for the customer and a pain for the vendor as well. Because the vendors put lots of effort to build connectors to delight their customers and allow them to build the project they want to build.
“But they never seem to have all the ones that the customer needs,” he continues. “The customer says ‘Have you got these 15?’ The vendor says, ‘I’ve got 12.’ And their ability to innovate is then curtailed.”
Scullion says this situation can be alleviated somewhat with the universal connector, which is included in the Matillion ETL product. The universal connector allows users to build their own connectors in a drag-and-drop manner. No coding skills are necessary, Scullion says, and the resulting connector is published as a REST API.
The universal connector is only available in the Matillion ETL offering, subscriptions for which are priced on an hourly basis, with an introductory setup costing around $15,000 per year for a connection to a single data warehouse. The company also offers a free Matillion Data Loader product that does not offer all of the enterprise capabilities of a full-fledged ETL tool. It sports about 40 pre-built connectors; the Db2 for i connector is not one of them.