Can’t spell Actian without AI

Not everyone stuck around when Ingres, the database services company Emma McGrattan worked for back in the early ‘90s, was acquired by Computer Associates. But she did.

Thirty years later, McGrattan is still at that fully managed database and data analytics company — since renamed Actian — where she is senior vice president of Engineering. Following a succession of ownership changeovers, Actian now is wholly owned by equity firm HCL Technologies.

Protocol chatted with McGrattan about what the company is looking for in new partnerships and possible acquisitions to help fill in gaps. She shared why Actian chose to build its own data quality product rather than partnering for one. And she talked about why she’s surprised — and disappointed — that more customers haven’t moved to the cloud more quickly.

This interview has been edited for length and clarity. A longer version can be found here.

What are Actian’s customers looking for from you when it comes to assisting in their AI and machine-learning-related projects?

So AI is definitely an area that we put on our horizon suite, an area that we're really interested in. We're probably going to do some acquisitions. There's lots of small companies that are struggling in the area of AI and ML, and we think to really bridge some gaps that we have, and the technology, that we’ll do an acquisition or two.

In the area of AI and ML, we don't have a huge presence today. But we have recognized that as you deliver a platform, it's one of those use cases that as people are dealing with large volumes of data, it doesn't matter what the industry is: The AI and ML [become] interesting.

So when we're talking to, let's say, an insurance company, and they're looking for fraud, DataRobot’s interesting, H2O.ai is interesting. But I think, given the heft of HCL and their deep pockets, that it's likely that a couple of acquisitions will help us fill in the gaps.

You're probably getting pitched constantly with all sorts of technologies and companies that want to partner with you. What are you evaluating right now?

Data governance is one area that we're really interested in partnering as well.

And then we're building our own data quality solution. We had previously thought we’d do a data quality partnership, but we've decided that data quality is so integral to the platform that it really is something that needs to be within our control.

But a data catalog is something that we could look to partner for. Our customers want to know that there are curated data sets within the warehouse that they can trust. And can you publish a list of those data sets and provide the provenance of who's curated it, and almost have, like, a Yelp rating system where you can say, “I'll give this five stars.”

We do today provide the ability to reach into a data lake and to pull data from a data lake into the warehouse, either to persist it as tables within the warehouse, or just reach out for it in the context of satisfying a specific query. And there, I think things get much murkier. When you're dealing with data in a lake, providing some level of confidence that that data has been curated, that you can trust the data, I think it's important.

Figuring out what are the missing pieces there so that we can satisfy the needs of the data scientist is important to us. And that's an initiative for 2023. So, we're right now figuring out: How do we deliver on that [in] ‘23 by bridging that gap with an acquisition? And it may be that it starts life as a partnership, and then, as we build more knowledge of the organization on the technology, then that could turn into an acquisition target.

Actian has been around since long before cloud. Do you still have a lot of customers that are using your services on-premises?

Surprisingly, yeah. We deliver a fully managed service, so you don't have to have people that know the technology, you trust us to do that. So that means that you don't have to have trained experts available 24/7 to support your data infrastructure: We provide it.

A lot of our customers have peaky workloads [i.e., data processing workloads that rise and fall based on various business factors]. The Irish Revenue Commissioners, which is the IRS equivalent in Ireland, run on our software, and coming up to tax day, they just can't get enough resources to throw at this problem. So cloud is perfect for that kind of workload where it's peaky.

Retail is another space in which, obviously coming up to the holidays, they see massive peaks, and cloud gives you that flexibility to grow the environment as you need it and then shrink it back. And you don't have to buy and build for peak. So I think cloud is attractive — the cloud economics make a lot of sense.

We've all moved to working in remote environments and working from home. I want to be able to work from any place. I’d like to take the ferry to Fire Island and work on the beach over there. Battling through VPNs and all of that I think is [not] interesting anymore. I think we need to make it convenient to access data, and cloud enables that.

I’m kind of disappointed that we don't see more customers move to cloud work quickly. It may be that a lot of our customers are on a journey to cloud; they're just not there yet. Hardware refreshes or making decisions about shutting down data centers might be the catalyst for them to look at cloud, but they’re definitely slower to adopt it than I’d like.

A MESSAGE FROM LOGITECH

In the New Logic of Work, solutions meet people where they are now, not where they were last year or even last week. And it means providing tools and experiences tailored to different people and roles in an organization. At Logitech, our products create opportunities for people and organizations to thrive.

Learn more

Jacinda Ardern has seen enough

New Zealand Prime Minister Jacinda Ardern’s Harvard commencement speech on gun control and protecting democracy resonated beyond the Cambridge, Massachusetts, campus, coming just two days after the Texas mass shooting that killed 19 elementary school students and two teachers.

Ardern talked about how New Zealand quickly enacted gun reform following the 2019 terrorist attack on two mosques that killed 51 people. A commission found that the terrorist, who livestreamed the attacks on social media, was radicalized online.

“We knew we needed significant gun reform, and so that is what we did,” Ardern said. “But we also knew that if we wanted genuine solutions to the issue of violent extremism online, it would take government, civil society and the tech companies themselves to change the landscape.”

While much has changed as a result, important things haven’t, according to Ardern, who called on social media companies and other online providers to “recognize their power and act on it.”

“That means upholding their own basic terms of service,” Ardern said. “That means recognizing the role they play in constantly curating and shaping the online environments that we’re in. That algorithmic processes make choices and decisions for us – what we see and where we are directed – and that at best, this means the user experience is personalized and at worst, it means it can be radicalized.”

There is an urgent need for responsible algorithm development and deployment, Ardern said.

“We have the forums for online providers and social media companies to work on these issues alongside civil society and governments, and we have every reason to do it,” she said. “Let’s start with transparency in how algorithmic processes work and the outcomes they deliver. But let’s finish with a shared approach to responsible algorithms — because the time has come.”

Facebook has a massive data governance problem

When a leaked document revealed Meta’s own admissions that it had little control over data flowing through its ads systems, lawmakers complained and privacy advocates sighed a collective “told you so.”

A privacy conundrum? Sure. But the leaked document exposed what many enterprises grapple with today: a major data governance gap. “This all points to an overall lack of governance,” said Mark Donatelli, managing partner at marketing data technology consultancy Cimply, after reviewing the 2021 document, first reported by Vice’s Motherboard.

Meta revealed it had “140 data processing systems” and “tens-of-thousands of uncontrolled data ingestion points.” Ads data engineers said the company needed to attach policies governing data from those ingestion points like APIs, and ensure that as data passes from one platform to another, those policies don’t get dropped.

They proposed a plan to rewrite and migrate those unwieldy data sources into controlled choke points where annotations showing data lineage — where it comes from, how it’s generated and what policies apply — would be added.

However, “at no point did they consider or recommend any solution involving removing access or blocking end points,” Donatelli said.

But as Meta CEO Mark Zuckerberg famously reminded Utah Sen. Orrin Hatch in 2018, Facebook makes money by running ads. Thus, Facebook did not want to turn off the data spigots that fed algorithmic models that make its ad system one of the most valued by advertisers in the world.

“They could have turned off all non-Facebook offsite endpoints and reduced data access by 60%,” Donatelli said, but “a lot of models would break.”

Around the enterprise

Google Cloud doesn’t work with the oil and gas division of Saudi Aramco,according to CEO Thomas Kurian, pushing back on critics that pointed out that its contract with the world’s biggest oil company isn’t exactly a feather in the cloud provider’s sustainability cap.

A MESSAGE FROM LOGITECH

We help individuals and teams feel seen and heard regardless of location, so they can do their best work, their way. Our solutions comprise a complete ecosystem of video conferencing hardware, software, services, and world-class partnerships. From boardrooms to living rooms, hot desks to huddle spaces—we’ve got you covered.

Learn more