The reality for an enterprise with lots of customers in different geographies is that in some regions and for some verticals it’s fine to store information in the cloud – but for others, it will need to be managed in a different way.
For reasons of security or data governance, for instance, some assets might reside in private cloud or a data centre, with many businesses taking a hybrid approach to cloud adoption depending on their needs.
A recent survey carried out by data management system vendor Cloudera found that more than two-thirds (68%) of the 850 IT decision makers polled currently store data in a hybrid environment – proving that hybrid cloud is becoming the de facto model for most organisations.
Yet almost three quarters (72%) of these IT leaders found that extracting value from data sitting across multiple cloud and on-premises environments was becoming increasingly complex with assets stored in different, inaccessible silos.
According to Chris Royles, EMEA field CTO Cloudera, key to cloud migration and data management is understand the workloads your business is dealing with.
Talking to TechInformed at the company’s annual client event Evolve23, he explains: “What we’re finding is that some organisations have stalled in their cloud migration. So, they’ll have a top-down directive to say, ‘we will move to cloud,’ but then they get stuck along that journey because of the complexity.”
“Our recommendation is understand the workloads you are running today, if you are able to modernise then do so,” Royles adds.
“Then, once you understand the workloads that you’re running, choose those that are most effective on public cloud; workloads that benefit from elasticity; workloads where volumes change – flash sales in retail for example.”
Another area where companies are finding cloud useful, Royles says, is for workload isolation.
“The general population within an enterprise might just want to do analytics and use data as part of their day job. Search some dashboard and makes choices. But that might be a different workload to the chief financial officer who is doing global roll outs and looking at quarterly cycles and load.”
For these organisations having easy access to specific data, no matter where it’s stored, is key.
LGIM’s approach
Creating a cloud-first data ecosystem that allows a firm to integrate data from different sources and streamline delivery to support business decision-making is the goal of Legal and General Investment Management (LGIM).
Appearing onstage at Evolve 2023, the tech and data teams at one of Europe’s largest asset managers shared their vision with delegates.
As a asset manager, data is fundamental to its business. And how different LGIM teams source the company’s data and present it to different organisational units within the business subsequently underpins its cloud and data strategy.
“Our main objective is to get the relevant data to our business and our clients faster. And we want the data to deliver value to the business,” says Wulstan Reeve, head of Data Marketplace, LGIM
However, Reeve acknowledges that pulling in the right data for people to access cam be a complex task and ideally, it also needs to be “frictionless” and accessible in the format people require.
Data integrity and accountability were the other requirements LGIM needed to factor in, he adds.
Cloud-first
As part of the company’s tech strategy, it set out the capabilities required from a data platform which, according to LGIM’s senior enterprise architect Stuart Toll, was all about being cloud-first.
“Particularly for an asset management firm, we’ve got lots of data coming in at different times, we need to curate data at different times, we don’t’ want infrastructure running 24/7,” explains Toll.
“So, we need the agility of being able to deliver on demand and to tear it down once we’re done for cost effectiveness,” he adds.
Time to market, integration time and skills were the other factors used to assess a short list of data platform providers.
“For some of them, we were concerned about the amount of additional integration needed and bringing other tools in to build up the ecosystem,” says Toll.
He also considered the skills needed to run the platform, adding “We didn’t want to be tied into any proprietary technology.”
While the company does adopt best-of-breed tools where needed, it selected Cloudera’s data platform which claims to support hybrid environments as well as multi-cloud unified security, governance and metadata management.
LGIM’s preference for cloud-first solutions included the implementation of Cloudera’s public cloud offering which it runs as part of its platform-as-a service.
“That meant we had to get comfortable with a vendor being able to stand out compute and software within our own Amazon account. But what that gave us was the capability and the agility and scalability we were looking for – plus to make sure that the data remained firmly under our control,” Toll notes.
It also meant that LGIM was not reliant on any one cloud vendor in terms of infrastructure – helpful as the firm runs a multi-cloud offering: workloads are primarily Amazon, but the company also uses Azure and other vendors.
Deliver value faster
Matt Bannock, LGIM’s head of data engineering, described how the firm used a centralised Data Mesh framework to manage its data.
Historically different parts of the business sourced their own data, copying it onto their platforms from different systems – which often involved “lots of duplication and effort” – not to mention a bottleneck of requests for the IT department.
By centralising data into a data warehouse, a mesh enabled the business’s SMEs to use that data and then generate their own insights.
“They are the people that understand how to use their data best, rather than coming to IT and saying ‘can you build this for us?’ , we want to empower all the SMEs in the business , the investment quants and finance people so they have the data available to them – and they can build out their own insights, their own dashboards etc,” says Bannock.
“The platform we’ve chosen supports multiple tenancy logical separation of domains all on the same cloud infrastructure,” adds Bannock,
He adds that this is important because the project required them to balance technology with delivery.
“And we need to build on our strength and LG’s strength which is creating funds that deliver strong returns for investors. Our time and money are much better spent getting value from our data than it is string lots of different tools together,” says Bannock.
He explains that the mesh provides a “central but local” experience allowing users to pull data into a local work area and start working with it as if they had their own data environment.
Reeve expands that, they built out the core of this data fabric by partnering with the business on high-priority use cases that would “add value, solve business problems, and enable us to incrementally move towards our vision.”
He adds: “We partnered with our Investment Risk team, for instance, and worked with them to make risk profiles of portfolios more transparent and have more capabilities so that they could have more engaging conversations with their clients.”
“Similarly, Quants are also familiar with data – they have that data mindset which makes it easier. With others we will have to support them on that journey – but as they begin to see the value from other use cases, they will want to joint the party.”
According to Reeve, LGIM is still very much on the journey, but the endgame is for the business to be able to self serve the data it needs without needing to go through the IT department.
And Reeve ends the presentation with a note of warning: Data transformations are not for faint hearted. They are best, he says, taken in “little evolutionary steps that add incremental value to the business and help you learn as you go along.”
“Try and push it too fast and it will feel like a revolution,” he warns, “and with revolutions someone always ends up getting hurt, usually people that it’s been designed to help, and there will be blood on the carpet.”
Subscribe to our Editor's weekly newsletter
Share
