The IT industry is defined by change – both the new technologies and the impact that these have on business. In the CUBE Conversation video below, Brian Gracely and I provide analysis on the shifting infrastructure stack and what limitations or gaps need to be taken under consideration.
Many of the technologies discussed here are about helping businesses go faster and when it comes to storage, flash is (mostly) about speed. Not only the speed of storage, but also how that directly impacts business productivity. Flash came back into the enterprise IT picture in 2008 – Wikibon believes that we will see more innovation with new flash architectures in the next 4 years than we saw in the last 6. Flash technology is seeing rapid adoption with hybrid and all-flash arrays (AFA), another wave of solutions with even higher performance and lower latency is coming with Flash as Memory Extension (FaME).
Reality: Customers are well past asking “why” flash, although there is not full comprehension as to the full extent of leveraging flash and the impact on applications. Flash is becoming prevalent in large public clouds, including more all-SSD offerings from AWS and Azure (Wikibon captures this in the hyperscale Server SAN discussion). The vision of the all-flash data center is good, there will still be a robust mix of performance and capacity (see Wikibon’s Latency vs Capacity Storage Projections). Reminder that there is no such thing as removing a bottleneck, you just move it somewhere else (and if it isn’t a component, check the people).
Hyperconvergence / Server SAN
While flash can help the speed and agility of IT, the biggest opportunity to help IT is to simplify the delivery and operations of infrastructure. Converged and hyperconverged infrastructure are both part of the trend of users seeking simplicity by buying fully integrated and tested solutions. These options are much simpler not only to deploy, but hold the promise of moving the needle on how much resources are spent on operations – industry data shows that IT spends over 75% of budget “keeping the lights on”; this statistic has not changed in the last decade. Wikibon’s focus on Server SAN points out that while today, the primary deployment methodology is as an appliance, these solutions are defined by the software, as proven in hyperscale environments. Nutanix (Dell OEM) and SimpliVity (Cisco partnership) have seen growth of the software offering, and it is the primary focus of Maxta and ScaleIO; note that ScaleIO can also create separate pools of compute and storage so that the solution is not “fully hyperconverged”. The value of these solutions is that they create pools of storage rather than traditional arrays. New architecture will have a very disruptive impact of traditional storage architecture sales over the next 10 years.
Reality: Users need to look at the applications, while mid-range customers can sweep the floor with these solutions, in the enterprise today they are typically deployed as a project or part of the environment, not all (so beware that it does not become yet another silo).
VMworld update: For the second year, hyperconvergence was a big buzz of the event. VMware Virtual SAN has made a lot of progress despite signiciant FUD that the product was struggling or that EVO:RAIL is dead (VMware states over 2k VSAN customers and introduced the EVO SDDC product). Customers are likely to sign up for an ELA when they purchase VSAN (VMworld customer panel on theCUBE). Additional VMworld interviews on theCUBE (see all videos) included VMware Storage GM Charles Fan, VMware Storage CTO Christos Karamanolis, Nutanix CEO Dheeraj Pandey, Hallmark Business Connection (Nutanix customer), and the CEOs of SpringPath, Pivot3, and DataCore.
OpenStack has won the marketing war for customers that are looking at building an on-premises cloud solution. OpenStack is built using open source and there is the promise of less “lock-in”. When users look at building a private cloud, the top options are VMware and OpenStack (note that OpenStack is not a single solution or product, and VMware offers VMware Integrated OpenStack (VIO)). There’s been significant consolidation (see OpenStack enters its second epoch), so there is now an opportunity to get beyond the hype/buzz that characterized OpenStack for a few years, and focus on improving operational efficiencies. It is worth noting that OpenStack is looking at replacing/updating a very deep technology stack, after only 5 years, there are already a number of useful open source tools. My summary of OpenStack is that like Linux, it will become a critical component of the stack and will be especially helpful as an integration engine for new technology adoption.
Reality: There are still too many projects in various states of maturity and comparing various OpenStack solutions/integrations is very difficult. The OpenStack Foundation is looking to address both of these issues. A number of companies are helping to make OpenStack a managed private cloud (on or off premises) including Cisco (Metacloud), IBM (BlueBox), Platform9.
Docker is leading the industry (hyper) interest in containers – there have been over 500M downloads in the last year. It is still very early in the maturation of containers, the evolving container architecture stack is quite complex. Maturation of the technology stack includes updates and extensions into storage and networking. In addition to Docker, there is growing interest in other vendors including CoreOS, Hashicorp, and Kubernetes. One of the most impressive things about the container marketplace is that it is very much a community centric solution; this is similar to VMware a decade ago, but maturing and growing even faster. There is a large ecosystem rallying around Docker and containers in general. Of note, Microsoft is moving very fast to support linux containers in Azure and create Windows containers (see Microsoft August 2015 container update).
The recently announced (originally Open Container Project) Open Container Initiative (OCI) enables runC as a consistent standard that platform vendors can leverage.
Reality: While containers are leveraged heavily in many large cloud deployments, they are still rare in production at scale in enterprise companies (Shopify shared what it took to get Docker into production on theCUBE at DockerCon 2015). Security is still the most referenced red flag for deploying containers, persistent storage and advanced networking are still in the early days.
VMworld update: See Brian Gracely’s research on vSphere Integrated Containers and Photon Platform – Is VMware Building a DevOps Framework for the Masses?
Container 101 Wikibon Whiteboard video
Cloud Native Application Platforms
Platform as a Service (PaaS) has long been touted as the future of cloud computing and now most vendors stay away from calling themselves PaaS. The cloud native platform discussion is hot and has a new Foundation (cloud native container foundation (CNCF)) driving some standardization. See Cloud Native Application Plaforms – Structured and Unstructured, and Technical Dive into Cloud Native Application Platforms.
Reality: Platforms are a big change for companies to be able to adopt. PaaS is the smallest piece of overall public cloud revenue today; it has strong growth, but remain smaller than SaaS and IaaS.
Action Item: CIOs must leverage new technologies to grow and transform the business. The technologies listed above are creating significant changes to the vendor landscape, but none are a silver bullet to IT. Practitioners must have a solid understanding of its application portfolio and the needs of the business going forward.