Hyperconverged Infrastructure
Sharing new Wikibon Server SAN research and interactive Q&A

What is your current state of adoption of Server SAN/hyperconvergence?

4 years ago
#serversanScalable InfrastructureChat with thought leaders, analysts, and customers on new server/storage architecture
Dave Vellante
Q8. What's the relationship b/tween DevOps and ServerSAN and what are the organizational considerations that warrant attention
Steve Kaplan We're seeing a great DevOps use case for HCI + Acropolis #ServerSAN
Julian Wood getting more infrastructure out of the way to deploy apps, simpler API provisioning rather than to multiple systems.
Dave Vellante For those who don't know...Acropolis is the Nutanix platform that essentially allows you to connect clouds and move workloads where necessary
Julian Wood problem is HCI still expensive today so DevOps being prefered in public cloud with simpler API than internally
Rich Holmes programmers are spending the majority of their time statistically developing for rest. That means a solution that supports rest API is going to be critical for easy development.
Keith Townsend depends on what direction of DevOps you are coming from? Developer driven vs. infrastructure driving. I don't believe it's commonly a joint equal movement within orgs.
Dave Vellante Acropolis received a lot of interest at .Next this year from practitioners that wanted to reduce reliance on vmware
Brian Gracely simplifying the infrastructure (more consistent) definitely helps the automation elements of DevOps
Keith Townsend ServerSan removes silos. DevOps is positioned to remove or lessen the impact of silos. Seem complimentary.
Brian Gracely for example, you rarely hear DevOps discussions that involve separate SANs. in fact, most call them out as a major bottleneck
Steve Kaplan I personally haven't seen that per se as a driver, but rather a desire for greater simplicity, scalability & security
Brian Gracely if you look at the 2015 State of DevOps Report, Ops is where most of the cost-savings & consistency comes from (API driven, automated, using config-mgmt)
Thomas Bryant What is the driver/reasoning behind that reliance?
Gabriel Chapman if your platform doesnt have a strong API capability, it simply will not be relevant especially within true DevOps focused orgs.
Keith Townsend
I've had customers say that server and storage cost is a rounding error for some of their projects. What's the value of the relatively unknown HCI?
Dave Vellante It's still 10% of a $1Tr IT market
Dave Vellante but your point underscores the shift in perception from one of infrastructure costs versus application value
Keith Townsend So, I guess my question is does the high-end need to embrace ServerSan to reach a tipping point.
Steve Kaplan Yup, HCI is just the starting spot for revolutionizing IT. Value = simplicity, scalability, resilience, productivity, slashed downtime - & lower cost. #ServerSAN
Keith Townsend I hear this downtime comment a bit. Where's that rationale.
Dave Vellante are you saying infrastructure doesn't really matter?
Keith Townsend Nutanix for example is a ~4 year old company. If I've received 5 9's uptime with my legacy environment how do I justify the perceived risk of HCI?
Steve Kaplan Check out hot-off-the-press IDC study showing ave 98% decline in unplanned downtime & 100% decline in planned downtime #ServerSAN
Rich Holmes The savings comes from rapid deployment. Being a post sales engineer myself in the past my experience has been when there is an issue with high-availability it's generally a human mistake and these solutions often mitigate that factor.
Keith Townsend @Rich_A_Holmes that's a good argument. Worrying that a complex infra is setup as designed keeps me up a night. (Not really but I worry about it).
Rich Holmes Not to mention the time it takes to test a solution. If you get a solution that fully integrate into the hypervisor's management systems from VMware and Microsoft then testing both destructive and nondestructive is quite simplistic. @CTOAdvisor
Steve Kaplan
Earlier in year, we had Fortune 200 company start out with 7-figure PO for Oracle & SAP on Nutanix. Now they are expanding to less mission-critical workloads.
Keith Townsend 7 figure PO for Oracle & SAP isn't that big.
Keith Townsend but it's good movement.
Rich Holmes this is similar to early adoption in server virtualization
Keith Townsend Depending on the criticality of the SAP environment I know Fortune 500's that spend mid-to-high 8 figures on SAP infrastructure.
Gabriel Chapman I shudder at the thought of the licensing costs
Rich Holmes people often overlook the fact that Microsoft does not support clustering on NFS based systems and hardware HA is only 1/2 the battle.
Keith Townsend Yes. Hardware is a very small portion of the cost. Almost an after thought.
Rich Holmes the demand is increasing for these solutions to offer true enterprise functionality
Rich Holmes all inclusive licensing based on capacity is the answer to keeping licensing costs down. Particularly as we are using industry-standard servers for this technology and the amount of processors in Cores change rapidly
Rich Holmes capacity is the only fair thing to measure because it is relatively within the customers control.
Dave Vellante
Q5. Lots of questions about the importance of infrastructure - does it matter?
Stuart Miniman Discussion of HCI isn't about "cheaper" it is about "simpler" - want to get rid of undifferentiated heavy lifting and allow IT to focus on adding value to business
Dave Vellante @floyer says storage spend is not a rounding error - rather it's an enabler of value creation
Keith Townsend when the risk of changing is worth it. You have to prove that a shift in infrastructure is worth the cost (risk) of shifting. Is the reduction of latency offer that much more business value? I'm not saying yes or no. That's the consideration.
Keith Townsend @stu companies care about simpler when there's tangible value. Application owners don't care about simple infrastructure unless it results in less downtime and agility. Does ServerSan bring these? If so, it will prosper.
Jim McBride No - companies need to stop accepting/designing needlessly complex solutions in this space and focus on normalizing it - and driving revenue-generating applications.
Rich Holmes I would say the infrastructure can matter for example better hardware can mean things such as sensors that can relay proactive failure measures and feed of that information into VMware or Microsoft avoiding failures to begin with.
Gabriel Chapman @Rich_A_Holmes apps determine the relevance and value of the infrastructure. IE Oracle/SAP vs Twitter. One of those requires validation, one requires engineering.
Dave Vellante
Q9. What's a storage administrator to do?
Dave Vellante "Get some training!"
Keith Townsend Learn networking (SDN) because no one is converging networking :)
Rich Holmes just like platforms are converging many organizations are looking to converge storage and server teams. Choosing a solution that can be managed using a centralize solution such as VMware V Center means that learning curve is much easier.
Steve Kaplan Agree w/@dfloyer. Days of LUN carving and zoning are coming to an end
Thomas Bryant Stop trying to be a silo would be a good start. Storage needs to be a consultative role, work with business vs. black box
Gabriel Chapman I've been telling "storage admins" to learn something else for years. Not a popular thing to hear at a storage conference, but its the truth. Collapsing silos and being part of a cross-functional team will become the norm eventually.
Brian Suhr HCI is really only option converging the skills today. Storage teams should learn new skills to move up the stack.
Thomas Bryant @Bacon_Is_King Role Role will never die, just fewer slots for dedicated person/team. SW can do a better job for the most part at their role than a human can.
Rich Holmes in fact the demand for smart people that know the environment will only increase as these solutions become prominent in remote office scenarios as well.