As it is late November, Parcel delivery company Hermes is gearing up for its busiest time of the year. In the run-up to Christmas, the company delivers double the number of packages it handles during the quietest days of summer. As such, Chris Render, the enterprise data services manager describes Hermes as a “peaky business”.
The 260 million packages it delivers annually are handled by 14,500 “lifestyle couriers,” and may pass through 4,500 ParcelShops. This all generates massive volumes of data with billions of data points coming from automated scanning devices, automated sorters and the thousands of couriers with hand-held scanners.
Render and his team of 45 data “multiskilled” data engineers are responsible for data infrastructure and data processing in the company and so looks after everything from traditional Oracle databases to SAP HANA in-memory databases.
“We wanted to modernise our data infrastructure."
In October 2017, Hermes executives decided to shift the emphasis of the company away from being parcel-centric and towards being customer-centric. “We wanted to modernise our data infrastructure and go through a true data transformation so that we could support a digital business transformation going forward. The other decision was to rearchitect how we handle data and that starts with a definition of a new enterprise data model,” said Render.
In the past, problems and issues in the delivery network could take up to 12 hours to be uncovered. This was because they were using classic batch ETL that would select data from a core transactional system which would be pulled through four times a day. Then reports, that take approximately an hour, would be run on those batches.
“So you’ve got the time of the ETL, you’ve got the time of the report to run and then you’ve got the latency before ETL, so you’re looking at significant delays for that data being available,” explained Render. This was clearly at odds with Hermes’ desire to be a real-time organisation. Now Hermes is able to intervene and rectify issues – such as a trailer with 10,000 parcels on board sitting in a yard for too long - that could cause packages to be delayed before they impact customers.
Hermes has moved data to the cloud and has a large set of databases in Oracle’s cloud, which Render describes as being new because not many people in the industry have done that yet. Hermes is now using software integration platform Talend as its major ETL tool. Render said: “We originally invested in Talend was as part of a new modern data platform so moving away from a traditional Oracle data warehouse into a real-time, in-memory data platform. Streaming of data was key so we invested in Talend as part of that platform.”
In addition, SAP HANA is the core database for the new modern data platform and is being used as a hot data store, holding transactional data for 40 days. Once the data is moved out of there, it goes into SAP IQ, formerly known as Sybase IQ. Render said: “We don’t want stale ageing data in there so that is supporting that real-time activity. If we want to do reporting over time we can basically bridge those two databases and produce trend reports.”
Hermes wants to be the first choice for customers when it comes to delivery companies and to reach that position, the company wants to have a 360- degree view of the customer to be able to give them better service. Render stated: “The data transformation is underpinned by everything we want to do as an organisation in putting the customer first.”