I’m currently going through the interview process for a VMware Architect position at a large company. Last week’s initial interview with the manager went extremely well and a technical interview was quickly scheduled the for this week (tomorrow to be precise). In preparation, I wracked my brain to try and guess what type of questions I would ask a prospective candidate for such an interview (for both a technical admin role and architect role). Having worked in the IT industry for over two decades, I’ve found that the people who fill such a positions are different in experience and even personalities. So, somewhere between “being curious to find a way to weed out those not qualified” and “just being an evil person to watch them squirm,” I decided that a graduated system would be in order.
Initial questions (generic):
- Can you give me your thoughts on virtualization?Admin role: While there should be no “bad answer” to this, there’s one major thing to listen for: excitement for the technology. Virtualization is a rapidly evolving and massively important part of the infrastructure. Obviously an interviewer would want to hear about leveraging the technology for consolidation ratios, reduced capital expenses, and doing more with less. As an virtualization administrator, they should truly live in the vSphere client/web client.
Architect role: Again, listen for the love of the technology because if they don’t have a love for it, you’ll end up with a stagnant design and one that may not leverage features that propel the business forward. They should enjoy reading case studies, look for deeper understanding of related technologies (storage, networking, guest OSs, security) as well as the business side of the organization (business processes, ITIL, microeconomics).
- How do you keep up with the changing technology in the field?Admin role: Look for examples tactical work and reading of blogs. For instance, admins should have a lab that they use to try out various work – usually at their home. They would be pushing for such a sandbox in the work environment, too (and hopefully the company will see the value of such a setup). If they say the have a lab, ask what the last couple tests they performed on their own time. A good answer would be a newer technology or trying to recreate an issue for troubleshooting; but they should give some better details in their answers. Most admins will keep up to date with reading blogs, VMware.com website, or attend VMware Users Groups (VMUGs).
Architect role: Here you’re looking for a much deeper and real-time engagement to the field. Most serious architects will have a home lab – some dedicated storage system, networking boxes, etc. Nested labs are usable and very common with today’s newer processors (where you can nest 64-bit OSs multiple layers deep). Staying up-to-date is different for an architect. In addition to the blogs, vendor sites, and VMUGs; they will usually use social media to maintain a pulse to daily developments in their area of focus and related areas. Many will have a personal account and a professional account for tools like Twitter – where the professional account is used to follow and tweet relevant items and keep their personal posts separate. As they have time, you’ll see them engage in the community and perhaps have their own blog. Finally, they will look for ways to improve the association between tech and the business.
- Provide an example of how you have used virtualization to address a specific problem or need in an organization?Admin role: Hopefully an admin will see their work as it affects the business operations (some do, some don’t). It’s been my experience that they will lean towards the tactical implementation of technology. So, while everyone’s answer will be different, you’d be looking for a clear example of a thought process (not just the spiel on work they’ve done). “Well, we had a problem with <insert problem> and the typical work-arounds weren’t meeting the needs of the <insert business unit> and so I was able to deploy <cool virtualization solution> and then the users were able to continue to do their work.
In the above statement, there’s a clear line between a problem or need, how it affected operations, and then a solution which had a positive outcome. If a candidate can convey this type of thought process, then I would want them on my team.
Architect role: What you’re expecting from an architect is a bit different. You want an higher-level and wider scope in the answer. Here you should expect to go beyond the tactical and into the “bigger picture” where you incorporate those sister technologies mentioned earlier (the more seamless, the better). “Our company had one last hold-out for virtualization: the DMZ servers. Since we adopted a ‘virtualization first’ stance, it only took persuading the security analysts that we could actually provide a better security posture using virtual instances rather than physical servers. By using a mixture of Private VLANs, hard-zone storage partitioning, and hardened guest configurations, we exceeded our internal security requirements while saving the company $100,000 in CapEx for the next hardware refresh cycle while reducing two full-time equivalents for analyst time to maintain daily operations.”
You can see this answer merges the various facets of the datacenter as a whole and how it relates to solving a design problem. Sure, a virtualization architect may not know how to configure the SAN fabric to meet these needs – however, at a minimum, she should be able to understand the capabilities of such systems and how to communicate those into requirements for the storage team. Further, there’s a direct resultant benefit to the business; in this case, it’s both CapEx and OpEx.
Technical Analyst (admin) Questions:
- Can you explain how VMware High Availability works?I chose this question to lead the technical questioning for a reason. It is a good lead in question because all admins should know at least the basics of HA. Hopefully it catches the absolute novices right away. Most techs outside of the VMware world equate HA to “no outage” or “full redundancy” so the protected service doesn’t go down. If you get this answer, nicely complete your questions and move to the next candidate.
A real admin knows that HA can protect VMs by detecting a node failure and restarting the guests to surviving nodes (yeah, there’s an outage while the VMs boot back up). Other information provided such as HA Admission Control, Guest monitoring, network & datastore heartbeats, and slot sizes (bonus points for the last two).
- What is EVC and what benefits and detriments does it have?Perfect time for them to explain how a setting can be a double-edged sword. EVC allows hosts that have similar (but slightly different) processors to still share resources together in the cluster. They still have to be supported by VMware and they still need to be the same manufacturer (Intel or AMD). However, different families can comprise the cluster (actually pretty cool technology). So the benefit is clear: you can have older and newer models together; build out the cluster as budget permits; re-purpose some of that hardware you just cleared off by virtualizing the system. Downsides: EVC has to be set to the lowest common denominator of the cluster; it must be enabled BEFORE guests consume resources (or they have to be powered down); it restricts VMs from using the more advanced processor functionality if EVC is masking those features for the sake of compatibility.
Let’s say you have two environments to set up. The first is an economy-class development server farm, say 50 VMs on five hosts; the second is a production VDI for the sales team so we can centralize their data and not worry about laptops losing data. Give me a few points on how you would make different design choices between the two environments.
First, if they are in front of you, do they look worried or excited to explain their plan? You should know what to look for by now…
For the actual answer, there’s no set answer, but hopefully you would get a sense of their technology background and diversity. Economy servers could have specifications such as:
- reside on lower-end storage systems (NAS, iSCSI, etc.)
- use either no RAID or slower RAID-5 or -6 for some data protection at the cost of performance
- server VMs could be thin-provisioned for even more resource savings
- 1Gb Ethernet connections… perhaps a redundant link
The VDI is obviously more important and will have better technology backing it:
- much higher-end storage (redundant iSCSI or Fibre Channel)
- RAID-10 and perhaps SSD for a linked-clone setup
- fully redundant 1Gb/10Gb Ethernet data connectivity to physically separate switches
- enhanced security on the objects to prevent accidental deletions
- reservations on the VMs to ensure no resource contention
The questions above can give you a good baseline on a candidate. They are what I would personally start with as a member of a hiring committee as they show a baseline level of knowledge and proficiency. My experience in this realm is that good techs have a certain je ne sais quoi about them. They actually care about what they do – it’s not a job, not a paycheck… heck, it can even be beyond simple professionalism; but how can you measure that? Hopefully the questions serve as a discussion starter and you can move from the concrete to the subjective from your own experiences.
Shameless Plug: Would you benefit from having a knowledgeable consultant as part of your hiring team? Contact me for more information!