Modern businesses aim for excellence and search for robust and agile next-generation technologies that optimize their infrastructure and solve the storage challenges they face.

As we dive into the changing and evolving world of storage, who better to answer questions than Boyan Ivanov, CEO and founder of StorPool Storage.

StorPool Storage develops one of the most reliable and speedy primary storage platforms that public and private cloud builders use as the foundation for their clouds. Their team has experience working with various clients - Managed Service Providers, Hosting Service Providers, Cloud Service Providers, enterprises and SaaS vendors.

VMblog:  Let's talk about cloud implementations. What's the most common mistake organizations make when building a public or private cloud? 

Boyan Ivanov:  The first step when building a new public or private cloud...is actually to analyze the states - usage, build-up, and other metrics of the existing cloud. Since very few cloud projects are greenfield, companies should spend the time to understand their current users, needs, and goals and then use this information to right-size their new platform. 

If the case is that one is building a greenfield environment - then there should be a sprint to assess the key needs and metrics, which tie back to these needs. And only then look for the right technologies to deliver on the needed target KPIs. This would ensure that the new project is geared towards what matters, rather than just building a cool new cloud. 

All too often I see companies focusing on the technologies or cool new vendors that they want to utilize and give little thought to the workloads, applications, and users of the new cloud. This is just wrong. The user and the application should come first. 

Last, but not least - in greenfield projects, look for solutions that are best in class and can start small and grow big seamlessly. Many vendors and solutions claim they can do this, but when you dig deeper into actual technology capabilities - it often is the case that most promises were marketing slogans, rather than technological capabilities. 

VMblog:  What about MSPs? If an MSP is creating a cloud infrastructure for their end customers, what should their goals be? 

Ivanov:  MSPs have several specific issues - a shift from on-prem to cloud services, improving their operational efficiency and profit margins, changing customer demands, and staffing issues - would be some of the more acute ones. 

This means that they have to provide increasingly more sophisticated services, in order to compete with Hyperscaler or at least a mix of on-prem, hybrid- and public cloud solutions. This makes their business model even more complex and operationally expensive. 

Thus MSPs should be especially focused on working with vendors, which have deep expertise working with MSPs and deliver solid, scalable, yet versatile solutions. 

In the case of storage, for example, one should look for vendors who can deliver a versatile data storage platform, which can support multiple IT stacks (think VMware/KVM/Microsoft/containers), it should also cover several performance tiers and use cases. It should be able to start small, but scale both horizontally and vertically, and even adapt to different deployment scenarios in different locations across the world. 

MSPs should also seek to work with partners who would provide high-quality managed services to them as part of the offering - this would both improve their operating margins in the end and will address their staffing issues at the same time. 

Finally, since the business and technological landscape is changing so fast - MSPs have to select technologies that are future-proof. For example, using products that can support multiple technology stacks will allow them the flexibility to change directions, depending on how the market shifts. For example, on the storage side this will be a product that can support most software stacks - say VMware, Microsoft, KVM, bare metal, and containers. And if need be - due to market shifts, acquisitions or any other reason - MSPs can swiftly shift workloads and customers from one IT stack to another one, in order to adapt to the new realities. 

VMblog:  IT companies and DevOps teams are another category where performance is critical. What should these environments be concerned about? 

Ivanov:  At the core of performance stays latency. This is maybe the most misunderstood metric of performance and I have spoken about this a number of times. So focus on reducing your platform latency. 

It is especially important for IT and DevOps teams, because speed has a very direct time and monetary impact on their operations. 

If a legal system at a law firm is slow, this is painful. But if you have a development team of 100+ programmers who compile code and it takes them 3 hours a day, instead of 45 minutes - that is a huge waste of both time and money. Worse, it demotivates the key resource for that company's success - their people. 

VMblog:  Similarly in container deployments, storage needs to be fast and available. Any tips for architecting storage to support containers? 

Ivanov:  My experience with container storage is limited to larger-scale deployments with Kubernetes, which need to run on bare metal because of their performance requirements or the size of the deployment. So take my view with a grain of salt here. 

However what we see on the market for these larger, typically hybrid-cloud deployments is that users have both containerized and traditional virtualized or bare-metal apps running alongside the containerized applications. This boils down to two alternatives:

• either get a couple of storage systems - one for the containerized stack and one (or sometimes more) for the traditional stack, each tuned to best serve the given use case; or
• get a potent best-in-class SDS that can cover both traditional workloads, but also has good integration with the container platform of choice. At StorPool we're a representative of this second group of storage software products, with a CSI integration with Kubernetes, which has become the de facto standard for containerized environments.  

VMblog:  Why is the storage part of the stack so consistently challenging? 

Ivanov:  Storage, together with networking are the two most problematic pieces in the IT stack. This is due to their complexity and the intricate interdependencies that exist between the 3 main components of IT infrastructure - compute, network and storage. 

Compute (servers) are one of the most straightforward pieces. Complex in itself, still it is the most well-known piece, as historically applications were run on a single server. There is the largest body of knowledge in that area and it's self-dependent. 

When it comes to networks things become more complex, as it connects many servers and networks have myriad features and settings. And when remote shared storage like a SAN (Storage Area Network) or SDN (Software-Defined Network) is added - this adds complexity to an already complex IT stack. As a result troubleshooting storage issues is hard, especially in the cloud area, where server count and data volumes are skyrocketing and network complexity to connect one to the other is increasing in sophistication.

VMblog:  In StorPool's evangelism it seems you're primarily competing with expensive storage hardware like SANs or all-flash arrays. What's that conversation like, explaining to an organization that they can get the performance or resources they need without dedicated storage? 

Ivanov:  It is a generational change. StorPool is the latest generation of storage software that can replace million-dollar all-flash storage arrays, with software running on standard servers, alongside applications. 

"Software is eating the World" is the cliche, but it's actually true. Ten to fifteen years ago storage was unavoidably a hardware appliance. Today people want cloud-native applications, run by APIs, end-to-end automation, utilizing as little hardware as possible. Most hardware is already virtualized, so people think of hardware as software anyway. 

In this world storage and to a large extent networks are software-centric constructs, which are part of the magic application software stack - fluid, programmable, and seamlessly scalable. This is core to the Infrastructure as a Service (IaaS) business model. 

So shifting from hardware-centricity to software-centricity is the master trend, driving the adoption of so-called "Software-Defined Storage". And StorPool is a leading implementation of this concept, tailored for large and demanding IT users, needing block-level storage in particular. 

VMblog:  Certainly the scale of enterprise data has grown. What's the best approach to managing scale from the storage and performance perspective? 

Ivanov:  There are several ways to handle this. In the old world of SANs, users had to choose a trade-off between scale, cost, and performance. There was a considerable complexity when designing a Cloud or large IT infrastructure setup. In many cases, cloud builders had to choose between speed and scalability or speed and data management features. So they ended with 2 or 3 different product families from the same or in many cases different vendors, which delivered different price/scalability/performance/data management sections. 

Today these trade-offs are a thing of the past. The best-in-class SDS solutions can offer practically unlimited performance and match it to the needed scale or data management & feature set requirements of most users. All in a single data storage platform. 

VMblog:  Ten years ago we were just beginning to talk about the software-defined data center, and now it's matured and iterated additional categories like infrastructure as a service. Yet storage remains a foundational component of all of these iterations. How has StorPool managed to stay right at the head of this evolution for these past ten years? 

Ivanov:  It's paramount to focus on what is really important. What drives this evolution? It's the changing nature of business demands and use-cases. 

20 years ago IT was one of the many functions of a company. Today IT permeates everything a company does. Let's take a bank as an example - 20 years ago their branch network and top-notch staff and service were what counted the most. Today it's their mobile banking app and online banking portal. In other words - even banks have become IT companies. 

I'm giving an example with banks, but the same applies to pretty much any other traditional business. And for the newer digitally native businesses as SaaS companies - this is 100% the case. 

And all this is also driven by technological progress and changing user behavior. So the users of today want digital, self-service, always-on IT platforms. These are more convenient and open 24/7. 

Then the businesses aim to provide these. And to do that - one cannot use the technologies developed for a different era, like mainframe computing, or fiber channel SANs, in terms of storage.

So the infrastructure has to be fluid, programmable, scalable, and self-healing. It just happens to work best if it is relying on a minimal set of standard hardware - compute servers and network devices - and then all the intelligence is in the software layer. 

This is why we see an accelerated switch from traditional, now legacy IT infrastructure designs to Software-Defined DataCenter (SDDC) and IaaS/PaaS/SaaS designs. What we do with StorPool and our storage and networking peers is to deliver the tools - such as SDS (Software-Defined Storage) and SDN (Software-Defined Networking) products in order to fulfill that need. And we strive to be best-in-class at the SDS layer and enable and fulfill the needs of businesses and end-users in the best possible manner. 

##