Talend CEO - cloud data migration is taking off, but why hasn't multi-cloud lived up to its billing?

I'll bet I can't find a single enterprise that doesn't want to apply their data to get closer to their customers, serve them better - and boost revenues accordingly.

It's the doing that's hard. Companies are stumbling with the realities of data management, and how you turn IT into a business driver without busting budgets. 

That's why I'm keeping an eye on data plumbing vendors like Talend. In 2018, Talend's user conference made an impression on me. As I wrote then:

The Talend Connect 18 keynotes pushed a heady new phase in next-gen applications, where portable/cleansed data (via Talend) and portable apps (via containers and orchestration a la Docker and Kubernetes) allows IT shops to avoid futile fundraising by moving from (expensive) legacy IT to cost-effective, modern infrastructures. (Talend CEO - IT must get unstuck from a legacy cycle to turn data into a business asset).

A year and change later, I'm on the phone with Talend CEO Mike Tuchen, looking for a progress report - and how the Stitch acquisition is panning out. I doubt Talend likes my "data plumbing" shorthand, so how do they describe themselves? Talend bills themselves as "Modern ETL." They want to establish themselves as the cloud integration leader, with Talend Data Fabric providing a "single suite of cloud apps for data integration and data integrity."

Why are cloud data migrations surging?

Readers know that marketing lingo isn't my thing, but: I keep hearing Talend cited by customers doing super-interesting stuff, as in this Oracle IBVI story, one that features prominently at Oracle OpenWorld again this week. My first question for Tuchen: with all the directions you pushed at Talend Connect '18, from serverless to multi-cloud to Kubernetes, what is sticking to the wall? Tuchen:

Right now our cloud business is on fire. You got a preview of that back in 2018. But that business has grown well over 100 percent, and has been for quite a while.

Each quarter, we've been releasing to Wall Street each quarter, what percentage of our sales have been cloud versus everything else. When we first did that in Q3 last year, the number was 14 percent. We got a collective yawn saying, "Oh that's not very big." And we're like, "Yep, but watch this space." And the 14 percent turned to 25 percent the next quarter, and 36 percent, then 43 percent in the most recent quarter.

So what accounts for their customers' cloud data shifts? Tuchen credits the next-gen tooling that is gaining a foothold:

It's really driven by the much more powerful capabilities they're able to get in the cloud with. You talked about serverless and Kubernetes and so on, on the processing side. But the machine learning capabilities they get, and the new cloud data warehouses are very, very impressive. That combination is super compelling.

Every customer we talk to is finding their path to cloud. That's driving this huge acceleration in our cloud business, and overall rotation across, I'd say, the entire tech landscape.

Does that explain Talend's November 2018 acquisition of Stitch, which aids Talend in providing "self-service" integration to cloud data warehouses?

Exactly. That was driven by the observation that it's gotten so much easier to build a cloud data warehouse right now, than what it used to be.

Building a data warehouse used to be a chore - to the point that "data warehouse" has a heavy-handed, legacy connotation. Tuchen says that's changing:

Back in the day, I used to work in a SQL Server business at Microsoft. Building a data warehouse was a year-long, or multi-year journey, that required a whole bunch of experts up front with a bunch of planning. Then you bought a bunch of hardware and did a bunch of heavy lifting.

Tuchen says that's changing:

Nowadays, an individual analyst in the sales and marketing team can spin up a data warehouse if they want to. It's that easy. And often times, that data warehouse, to do their first sales and marketing analytics, then grows up and becomes their primary data warehouse. Getting in on the ground floor, and being part of that really easy data warehouse building phenomenon was something that was really important to us. That's where Stitch has been playing a super role for us.

The gap between multi-cloud vision and reality

I get why Talend's cloud business would be surging. But what about the unrealized promise of multi-cloud? Multi-cloud was a big theme at Talend Connect 2018, but I still see a big gap between the vision of multi-cloud and the reality. The idea of customers being able to shift workloads between clouds for a variety of reasons, utilizing the best features of each cloud offering, and playing them against each other in negotiations versus single-vendor-lock-in, seems appealing. But in reality, I don't find customers are doing that in a very savvy way yet. What is Tuchen seeing?

Same. There are people that talk to us about that workload arbitration model or price arbitrage model. And maybe they use that during their negotiations, but in terms of on a daily basis, swapping stuff around, we don't see a bunch of that.

But Tuchen says almost all of Talend's large enterprise customers are indeed "multi-cloud," typically running clouds a team or division at a time. So what are the drivers for this iteration of multi-cloud?

• Divergent adoption - different divisions adopted different clouds - IT had to make it work. Tuchen: "By the time the central IT team started to pay attention, they already had a footprint in more than one cloud. Trying to uproot that and try to standardize it was too much work. So they said "Fine, the status quo is good enough, and let's just make sure that we can govern across that multi-cloud footprint."
• Lock-in avoidance - some customers are determined to pursue a multi-cloud approach to avoid the mistakes of the past. "We see companies that are doing this strategically. They're saying, 'We learned a lesson the hard way in the last go around, that being beholden to one large IT vendor was a painful experience. And so we're going to make sure that we have our bases covered and we have strategic independence. And so we're making sure that we have a multi-cloud footprint.'"

Makes sense - but the management of multiple data clouds is hardly frictionless. Can Talend help there?

Absolutely. Because for us, we can be totally agnostic about where you put your data, what you do to it, and we can give you a view of what's going on across your entire data landscape to help you govern it.

Tuchen is always handy with the buzzwords. He says customers are in need of this one in particular: the so-called "hybrid multi-cloud." In other words, data gravity may be shifting to the cloud for most, but it's shifting at a pace that suits. There's plenty of vital data in on-premise systems; customers expect Talend to help with that too.

Hybrid multi-cloud is a common term you'll hear as well, as companies still have an on-premise  data footprint, as well as whatever cloud footprints they have. Just keeping track of all that and being able to govern across that entire landscape is super-important.

The rise of Google, Oracle, and "specialized clouds"

Tuchen is seeing another multi-cloud trend: the rise of specialized clouds. So far, the cloud infrastructure race has been a head-to-head competition, with AWS out in front. But is that shifting? Tuchen:

Do we expect to see every cloud continue to cover all the bases and seeing a multi-front war across the entire stack, across those three vendors? Or will we start to see more specializations start to develop? I would argue we saw Oracle take a step in that direction. Oracle at one point had aspirations of being their own end-to-end cloud. And I think they're increasingly realizing that's not going to happen.

So now, they struck a deal with Microsoft to say, "You can run your application in Azure, but we'll have very high speed, dedicated connectivity between the Azure cloud and the Oracle cloud, so you can still run your Oracle database in the Oracle cloud." So they want to be the world's best, and in their world, preferably only place to run their Oracle database, but you can still run your application anywhere else you want.

Then there is the rise of Google:

Look what Google is doing with Anthos - Anthos being a Kubernetes orchestration engine that runs in a different cloud, or runs across clouds outside of Google's cloud. Given Google's strategic intent of winning in data, their combination of BigQuery, TensorFlow and Looker is a very strong combination, and they're seeing that as their primary thrust to win in.

Google's enterprise push is noticeable in the field :

We're seeing Google pop up a lot. If you'd asked me a year or two ago, Google is sort of a three to five percent player, and we're like, "Yep, great technology, we keep hearing good things about it, but no one's using it." That's demonstrably not the case anymore. They're moving fast from a small position. They're certainly playing from behind, but they're moving fast right now. So I think it could be real interesting to see how the next couple of years play out.

My take

I asked Tuchen if he's seeing the same customer surge in "buy over build" that led to our recent buy/build debate on diginomica. He says yes, but with a twist: he also sees vendors covering areas where customers used to have to build. Now they have a packaged option. What that comes down to is: more customer choice, rather than one absolute trend. Today's enterprises have more flexibility in how they approach IT (cost center, multi-cloud manager, builder of smart applications, data science/analytics governance, workflow automation, all of the above).

But there's still too much data friction. At Talend Connect 2018, Tuchen shared these stats: 40 percent of enterprise data has data integrity issues. And: 80 percent of an analyst's time is wasted preparing that data. Mix in data proliferation trends; I seriously doubt we've made much progress on those stats across the industry. As long as that data burden continues, IT is, by definition, more of a janitor scrubbing away than a business leader.

Can "AI" and the automation of data cleansing help? Tuchen definitely thinks so - but we agreed to table that part of our conversation for a couple of months. At that time, a major new Talend release should bring this conversation to a head.

There's also a conversation to be had around the rise of "serverless", containerization and Kubernetes - all areas Tuchen has highlighted in the past. There are economies of elastic scale here for customers, if you will allow me a buzzheavy phrase of my own. But: there's also a level of required skills - an IT evolution that customers need help with. That's a topic to revisit.