IT BS thread

With questions like those, I wonder if they really DO want to know how well you can BS something on the fly.

I'm sure you've had to dumb things down for non-technical folks before; in cases like those you have to be able to think fast and not insult someone's intelligence.

I know I have!

 

When I used to conduct interviews, I avoided those. People come prepared to answer strength/weakness questions and they always resulted in useless information. Randomized questions always told me how someone could think through a real situation.

 

I used to ask the greatest weakness question, but that was years ago. I quit after getting nothing but canned answers, deer in the headlight looks or being told "I have no weaknesses." (yeah right, I can tell you that the person with no "weaknesses" lacks humility.) In the end those kind of questions don't provide any useful information. As for BS'ing an answer I don't hire sales people so that's not a skill I want in my group.

 

We have a packet of technical questions that we've used for interviews for years. It has several in depth questions to figure out how or if the candidate can think through problems. A favorite question is about VMware ballooning.

But then one day I had a manager try to tell me that VMware ballooning was a good thing. He was quizzing someone who was about to go take the vcp exam. The guy couldn't answer the question so I very plainly said ,"ballooning is what happens right before the shit hits the fan. Ballooning is right before swapping and lots of calls from app teams about their servers hanging." he disagreed, and that's the difference between a manager and an engineer.

 

My favorite are the “my weakness is actually a strength” answers.

The manager or the engineer disagreed? I think I follow and I think we’re on the same page

 

The manager. The other engineer didn't know what ballooning was.

 

When I use to interview/tech screen candidates to join the VMware PS team, I would go very very simple with two questions to see if the VCP listed on their resume was a sham.

1.) What's the difference between High Availability and Fault Tolerance features in vSphere?
2.) Explain to me how a vMotion works and its dependencies

It was downright baffling sometimes how often VCP-holding candidates would fumble over those two. Typically after this point I would just ask about items on their resume.

 

I found it pretty common that most IT people don't know the difference between HA and FT.
HA isn't necessarily FT but FT includes HA.

I don't have my VCP and haven't used VMWare in awhile. Mostly Hyper-V for the past 5 years and at the new place we are running Nutanix AHV but ill give your questions a shot w/o Googling and let me know if I pass

1) HA allows for the rapid restoration of services but doesn't prevent an outage caused by things such as failed hardware. In contrast FT inherently is HA but allows for services to continue uninterrupted when there is a failure within the system. In the realm of VMWare an FT VM has two copies running. The primary VM and a hot standby VM on a different host. Memory changes are constantly synced from the active VM to the standby VM. If the active VM fails the standby VM is activated.

Additionally at the hardware layer redundant power supplies would be FT assuming they are connected to an A and B power grid serviced via different power grids/UPSs/etc. RAM would be FT if you are using mirroring. Both redundant power supplies and RAM mirroring are HA.

FT requires twice the resources that a single system would normally use.

2) vMotion works by essentially snapping the RAM state of the running VM and moving it to another host. Once the RAM state that was snapped is moved then the incremental changes in RAM from the time it was snapped to the time the new host has it is moved and the VM is "activated" on the new host. Since storage is "normally" shared between the two hosts only RAM state needs to be moved between hosts.

vCenter is required and a vMotion network must be defined. At one point vMotion itself required L2 connectivity between hosts but I don't know if newer versions can work over L3 routed networks.

 

Close enough and very good for not touching vSphere in many years and you answered much better than many that were actively applying to implement VMware at the professional services level, so I give them some scrutiny when asking fundamental questions at the core layer.

1. The HA and FT feature question is a little tricky as it asks about named vSphere features that share conceptual names with general IT terms. In short, HA is a feature enabled at the cluster level that will tolerate host failures, network isolation or APD/PDL states... typically VM's are restarted on living hosts depending on how you have it configured; however, the service interruption is minimal. Your assessment of FT was on target though.

2.) Solid answer on vMotion and I would have gave it a pass. The only item you were missing as a dependency is shared storage between the hosts for a traditional vMotion. Also you're correct on L2 connectivity; however you can vMotion over L3 now (but this wasn't the case 5 years ago)

 

I thought shared storage was implied and was thinking about adding to the answer but then I would have also gone into other scenarios like HCI where there isn't a traditional SAN/NAS. Hyper-V added a cool feature years ago with Shared Nothing Live Migrations which allowed you to move the VM + Storage to a new host/cluster w/o having shared storage. Obviously a lot more data is moved across the network when doing that and both hosts/clusters need to be online. It was great for infrastructure migrations from one cluster and hardware to another. Now Microsoft allows Mixed clusters where you can have 2012R2 and 2016 in the same cluster.

My last job I did an Active/Active Datacenter design and deployment that leveraged EMC vPlex at each location sitting in front off identical EMC VNX arrays. vPlex is a slick piece of technology allowing local servers to access local arrays but also having the LUN shared across sites. So you can have a shared datastore and a stretchered cluster between two locations. Other stuff was involved also for L2 LAN Extension, inbound path selection, etc.... I was pretty pleased once I got everything setup to see a VM migrated from one DC to another DC across town w/o missing a beat.

 

You know what? I'm a moron... Yes shared nothing vMotions exist in vSphere now too, things have changed since I had those pre-canned questions. I've been losing my low-level technical chops so bad ever since I moved to sales. My VCP's have long expired (and I'm sure I would struggle with some of that stuff now)... Fortunately my VCIX doesn't, but it doesn't say much when I forget stuff like that

Nice! I actually have a customer right now that is going through a VPLEX metro storage cluster implementation, being a fly on the wall for the engineering call have been a stark reminder of dying skills.

 

Yeah I had the same problem when I moved over to sales for awhile. Better opportunity came up last September so now I am Director of IT for a small SaaS company.
Fancy title, nice pay, staff of me and a desktop guy .

 

Nice, congrats! Eventually I’d like to do the same.

 

I sometimes wonder if I did the right thing by going into management. People problems are a lot harder than technical ones. The perk is the technical problems get harder too. When nobody else can solve the problem it lands on my desk to fix.

 

Ha .. management / non-management roles all the shit still landed on my desk to solve. In my last job I was one of the top engineers at the company and the only one who knew an entire stack from network (LAN/WAN/SAN), storage, servers, windows server, virtualization, voice, etc ....

On one hand its nice on the other hand you wonder wtf everyone else is doing.

 

VMs are fucking witchcraft. I only have to mention in passing that I need more servers for something, and some trolls in the dungeon magically shit them. Doesn't matter if they're Windows, RHEL 6, RHEL 7, AIX, Solaris, VMS, Novell, S5r4, or are needed in DEV, TEST, PERF, PREP, or PROD - instant servers. Like magic.

It's voodoo death magic, man.

 

It's the future, mannn

 

Sounds like IT these days mostly involves babysitting VMs.

 

As apposed to what?
Babysitting software running on baremetal hardware that no one wants to reboot because 1) it takes 30+ min and 2) no one is sure if the hardware will even come back up?

I'm not sure I would call todays IT babysitting.