Panelists: Rick Scherer, Duncan Epping, William Lam, Scott Lowe, Chad Sakac
Q: We know NSX solves big problems like Amazon datacenters. What is the smaller/smallest environment that it is deployed in?
A: Scott Lowe – In the smaller environment distributed firewalling is very, very powerful and a big use case for NSX. Its about finding the right business case. Chad- Still in the many 10s of ESXi hosts for the smallest NSX deployment.
Q: EVO RAIL seems like a VCE-like product but just a little bit more open.
A: Duncan – Built a hyperconverged platform that is very easy to use. VCE is typically multiple racks, but EVO RAIL is all contained in a 2U appliance. Chad – People will lump VCE/vBlock and EVO RAIL into the same bucket, and that’s a mistake. EVO RAIL is much simpler. EVO RAIL can scale up in smaller steps than vBlock.
Q: How do you compare EVO RAIL to other product
A: Chad – The EVO RAIL will use the same 2U appliances for the EMC devices. EVO RACK is an ECS with compute with generalized compute on top. You order from the OEM not VMware. Duncan – VMware is not shipping hardware, but rather defined a recipe with multiple OEM partners. Chad – EVO is not the same thing as EMC ECS even though the hardware is the same.
Q: EVO RAIL seems like a Nutanix competitor. What is the difference?
A: Chad – All are in the same category. The strength of EVO RAID has hypercoupled all the installing and licensing that is unmatched. For example, EMC will be adding Recovery point into their EVO RAIL offering. Duncan – Licenses are all included like vCenter, log insight, and vSphere licenses. One support contact number as well. Chad – EMC will lag the EVO release date a bit because they want to ensure global sparing, training, etc.
Q: How does Cisco ACI compete with NSX?
A: Scott – Every product as strengths. ACI may do somethings very well, like chip-level fabric latency. But most datacenters don’t suffer from problems that ACI may be trying to address. NSX believes that intelligence out to the edge and not into the core like ACI. NSX deeply integrates into other vSphere products.
Q: How can a customer obtain their VCDX?
A: Duncan – There are a lot of sample architecture guides out there that anyone can access. Templates can also make you lazy, so don’t worry about using VMware partner templates. Buddy up with other people going the VCDX route and learn from each other. Have your collegues look at your design to see if it’s correct or not and challenge your design. Rick – Get involved in things like vBrownBag. Scott – Find somebody that can mentor you and do mock defenses with, and challenge you. There are a ton of resources out there.
Q: ESXi has a lack of HA around APD events. The customer uses a script to help avoid the solution.
A: Duncan – The script mentioned on the forums is good, and he complimented the author. VMware is looking to solve the problem in product, at sometime in the future. VMware is aiming to solve the issue. Chad – There are cases when you should not respond immediately to a storage outage, so there’s no easy answer. William – Logs can be used to detect APDs and create an alarm on, and allow you to take proactive actions. Duncan – this is not an easy problem to solve and there are a ton of scenarios.
Q: VMware seems to be going towards hardware with the announcement of EVO RAIL.
A: Duncan – EVO RAIL is a recipe that enables partners to ship hardware that will be known to work well. It will be 2U, 4 nodes, flash devices, 10Gb NIC, etc. VMware is not in the hardware business. Chad – VMware is not selling software. Period.
Q: Does EVO RAIL use VSAN? And if so, how many nodes?
A: Chad – The 2U platform does use VSAN and the 2U module has four servers in it.
Q: Is there a method to programically create a VDS?
A: William – There are ways to create a VDS via script.
Q: Why won’t it scale larger than four appliances? Will there be multiple SKUs for difference licensing, like VDI?
A: Duncan – Testing has limited time and scalability is currently capped at a 16 node cluster. This constraint will be lifted in the near future when more testing is completed. VSAN is limited to 32 nodes at this point. EVO RAIL has a VDI target capacity of 250 desktops per 2U appliance. Chad – People use different size VMs, so there’s no hard coding of the VDI number for the nodes. It is customer specific. Each OEM can also vary amount of RAM as well. Expect multiple models from most EOMs.
Q: What is the best solution for micro companies that have 3-4 hours?
A: Scott – Email, do that in the cloud these days. Do the full cost analysis for each customer, for things like Office 365. Do not be afraid to move workloads to the cloud. Chad – With vCloud air there’s the new metered model which lowers the costs. Duncan – Do you want our customers go through the pain of migrations? Some customers want to really touch the hardware that their infrasturture on. Chad – The SaaS offerings are hard to compete with these days and are very attractive. Scott – Some people have feature of moving to the cloud, and it’s not all just about cost. But it is a viable option.
Q: How do the panelists “turn off”? they blog a lot, write books, etc.
A: Duncan – Buy your wife a lot of gifts. He sat his wife down and had a discussion about the time commitment. But the time invested in worth it. Chad – Says he loves what he does. Play is work. Chad can also burnout. Chad runs furiously then pauses, like working out and do non-work related tasks. It requires a conscious decision to not burn out. Duncan – Put your phone away and turn it off when you are with your wife. Turn it off…the biggest give you can give them.
Q: vCAC is moving to the cloud. Why?
A: Chad – vCAC is very complex so using it in the cloud removes complexity. Complexity is driving customers to the public cloud. There are some workloads that will remain on-prem. Scott – vCloud air is strategitcally important and will offering more software as a service.
Q: A lot of backup products use CBT for quick backups. Are there plans to help address CBT corruption?
A: Duncan – This is a known problem but engineering is working on it. But doesn’t know the exact status of the dev efforts. Chad – The first step to fixing the problem is recognizing.