As Splunk approaches the two year anniversary since launching Splunk Cloud, it is an opportune time to meet with SVP of cloud solutions, Marc Olesen, at the company's annual user conference in Las Vegasto get an update.
Despite being something of a 'new kid on the SaaS block', Splunk set out to make an impression that it is taking its cloud business seriously. So seriously that it is offering customers a 100% uptime service level agreement. That sounds unique among cloud providers.
You see a lot of 99.8%, 99.9% etc., but I don't recall a company say 'we can guarantee that there will be no outages or downtime'. One would assume that even if companies were very confident that they would never experience any downtime, they would also want to protect themselves legally with a 0.1% margin for error? AWS, which hosts Splunk's cloud offering, guarantees an uptime of 99.95%.
Can anything in technology be 100% certain? Olesen seems to think so. It's a convincing pitch. For one, Amazon Web Service's has effective failover strategies in place. Splunk is using its own product to monitor performance and incidents, which Olesen says gives him and his team additional assurances. Olesen explained why he isn't worried:
We feel really good about it and it's been 18 months since we had the SLA in place. We feel great about it for multiple reasons. We are running across the multiple availability zones, which provides a great deal of redundancy and full tolerance. Also our architecture is designed in a way that we provision a dedicated environment for our customer, which really helps us ramp up on the reliability side. Because in other situations or other architectures, one customer can compromise another customer's system performance. And then we are also, of course, using our own product.
Super confident indeed.
Olesen provided some detail about the company's cloud architecture:
We would describe it more as a multi-instance architecture, so each customer gets their own dedicated environment. It's software-as-a-service, but what we are finding and hearing from customers is that they want a solution that is instantly available and ready and one that they don't have to run, maintain, operate or upgrade. And they're not concerned about the architecture, as long as it meets their security requirements. So when we architected Splunk Cloud it was very intentional for us to go with a dedicated environment, because it has got a stronger security posture.
For a more detailed breakdown of the variations in cloud architecture, see my colleague Phil Wainewright's in-depth piece here.
Unfortunately, Olesen wasn't willing to provide a cloud customer count. All he could say was that over the past nine months the company has tripled its orders. I attended some sessions this week where customers were using the Splunk Cloud, but I get the impression that a significant majority are still focused on on-premise installations. That's to be expected, given that the offering is relatively new to the market.
Olesen was willing to say that Splunk's current cloud customer activity closely mirrors activity seen on-premise. For example, about 40% of the customer portfolio are using cloud to take advantage of Splunk's security offering, while a “sizeable chunk” are focusing on IT operations. He added that the use cases cover most verticals.
What about the company's decision to host its cloud offering on AWS. Not too long ago we picked up on disagreement in the industry about whether or not SaaS providers should be building out their own infrastructure, or whether it was okay to host on a third party provider such as Amazon. I'm inclined to argue that AWS is a sensible choice, but there are arguments for both. Olesen said:
The statement that I often share is a Gartner statement. Gartner said AWS is 10x the size of their next fourteen competitors combined. And I think it encapsulates a big reason why we have gone with AWS, they're such a dominant player and they're a great strategic partner for us. We've got a lot of go to market activities and synergies with them. They've got a really compelling infrastructure that they're constantly innovating on. It enables us to take advantage of that. They're doing it very effectively.
Finally, Olesen was keen to highlight that Splunk is not adopting a 'pro-cloud' stance when selling tocustomers. While we see other vendors trying to shift their business models towards cloud delivery, Olesen believes each customer will have a preference and that Splunk should cater to that preference. He argues that hybrid is likely to be the way forward for many.
We are excited that we are really the only solution in the marketplace with a seamless hybrid offering. The architecture we chose fits that as well, as we end up with feature parity between on-premise and cloud, and then of course they can then talk to each other.
There are different use cases [when considering cloud and on-premise]. There is certain data in certain organisations that they would not want to leave their data centre, so they have got Splunk Enterprise on-premise for that. And then for data that is less sensitive, they are sending that to the cloud. There are also capacity constraints. We have got customers that have Splunk Enterprise running successfully on-premise, but they are absolutely out of hardware or out of data centre space or out of the resource capacity to run in. And so then expanding capacity in the cloud is nice, but they still don't want that those separate or siloed environments.
But our approach to that is very much choice. We want customers to choose the option that makes the most sense for them. Our cloud offering is very compelling from a total cost of ownership standpoint and if it represents an option where a customer doesn't want to run or operate, that's a great fit. At the same time, if the customers have the expertise and have the desire to run it themselves, we absolutely want them to do that.