Having to perform under pressure.
I am sure we have all been there in one way or another and some of us handle it better than others. It doesn’t matter if the pressure if work-related, personal or a mix of both it’s still difficult and sometimes insurmountable to perform while under a great deal of pressure.
Recently I was asked to field a copious amount of questions regarding OpenStack, how we built our infrastructure, what worked, what didn’t, how did we implement Ceph, how did we deal with security, how did we overcome trials and tribulations of a data center transformation project. Not to mention while all of this was commencing we still were required to provide support, capacity planning, security, compliance, automation and continued standing up new infrastructure fault domains and provision workloads.
Now, I am sure some people look at this and say “Meh, I can do that. No big deal” but for me, my team, and those we interface with this was not such an easy task. You have a “changing of the guard” when it comes to deploying new technology. The internet of things as people are calling it or the software-defined data center, both terms which I absolutely dislike, are not a promise of the future, they are here, now, today, all around you.
There are some that respond well to change, they accept it, adopt it and love it. Those are the ones that excel. They are the first people to run into a burning building, diffuse the situation logically and then take action. There are others who call the fire department to come and put out the blaze and wait to see what happens. Lastly, there are those that sit around filming the building burn down without care and with disregard to those inside risking their lives by choice. I think most of us forget that firefighters choose to put their lives on the line every single day. I am by no means comparing information technology to firefighting so before you flame me, I am just using it an analogy.
I have always been a person who runs first into the fire. Sometimes this is a good thing and sometimes it’s a bad thing. Eventually, one of the times you run into a burning building you are bound to get hurt, trapped and/or needing assistance from the others around you to overcome overwhelming odds.
So, back to pressure, OpenStack and having to describe the good, bad and ugly pieces of it.
We run a fairly unique implementation of OpenStack as opposed to many others and do not segment storage from computing (we use cgroups to limit, account for, and isolate resource usage (CPU, memory, disk I/O, network, etc.) of a collection of processes for Nova and Ceph. Cisco UCS converged infrastructure was implemented as opposed to white box hardware and our Ceph implementation uses a combination of SSD and HDD for performance optimization (that’s actually pretty typical).
SolidFire is used on the backend for high-performance latency sensitive applications since it allows in-line deduplication, compression and offers out of the box replication. The diagram below is a good representation of our current implementation.
Data center transformations are an incredibly sensitive topic. Much more than most people would think and especially when you are moving to a new architecture that is still in its teenage years and growing faster than a weed. Getting people to comprehend OpenStack vs VMware, XEN or traditional KVM is already a tough task in some companies and once you start telling everyone that the DevOps culture is the wave of the future and “GUI’s are for sissies” some people get pissed off.
Once again, here is where the line gets drawn in the sand. Some people LOVE GUI’s and they refuse to use command line unless it’s absolutely necessary. I can tell you from my past experience doing consulting that I have seen more network engineers use Cisco Fabric Manager than I have seen use command line. Have you ever used Fabric Manager?! It’s a nightmare!
You begin to introduce your VMware-centric people to the simple Horizon dashboard and suddenly people begin to sweat and start asking questions such as, what storage is being used, how do I know where the VM (instance! its called an instance dang it!) lives? How would I troubleshoot a problem? Is there going to be a better GUI for me to use? What about something like the vCenter client, does OpenStack have that? Does OpenStack have HA? Does OpenStack have DRS (load balancing between hypervisors)? That’s when the situation gets ugly…..
Do individuals start to say “Who is going to support this? I’m not supporting this! There is nooo wayyyyyy I can support this. Are we sure this is the right move? I mean, VMware works well right. Why would we change?.” This is when people either perform under pressure or fall apart. I mean, imagine a room packed to the gills with IT staff all with this look on their face like you just called their Ferrari ugly and now you are going to trade in for what they view as a Pinto. My apologies to those that drive a Pinto. It’s a great car, no, really, it is……
That’s when you start talking about continuous integration, Containers, Jenkins, Git, Repos, version control, security, deploying bare metal as a service, deploying Hadoop as a service, deploying Containers as a service… pretty much-deploying ANYTHING as a service up to and including disaster recovery.
Shouldn’t it be as simple as a product group logging into a portal or catalog like Service Now and checking off a list of boxes, magically get an estimated operational price and then getting an email in a few hours saying their project is ready to go. The fact is, yes! Amazon has been doing it. Google has been doing it. Microsoft has been doing it. Actually, if you think about it all major hosting providers are doing just that. They write intelligent code and automation to provision software as a service.
I almost always lean towards software as a service over any other term. If you are providing a virtual instance, it’s running containers, it uses Ceph for the back end and Gluster for the shared file system, aren’t all of those software defined? What about deploying Hadoop on top of OpenStack via Sahara. Is that also not software defined?!
Companies of all shapes and sizes want the features of Amazon without the price tag and the ability to manage resources within their own private data center. It’s about knowing where you data is, how data is being backed up and saved, clear and concise monitoring of workloads, monitoring the data center specifications and statistics, validating compliance of customer data and securing it… it’s all about CONTROL.
This is what OpenStack provides. A set of tools for companies of all sizes to deploy a series of services and a standard set of APIs that allow developers, DevOps, and administrators to provision their own elastic scalable infrastructure in the same manner that Amazon does and in some ways BETTER than Amazon.
If Amazon was the latest and greatest thing since sliced bread everyone would be on the bandwagon, Facebook, Google, Yahoo, Microsoft and all others would just say “Screw this. Let’s just deploy into Amazon, fire 70% of our staff and we are good to go!” The fact is, Amazon has its own set of difficulties. Ever tried running containers in ECS? What about wanting to deploy in a region that’s not supported? Maybe governance requires data to not leave a country’s borders and Amazon doesn’t have anything in that area? What about risk and compliance?
Factually we have been through all of this before. Remember mainframes? I do. In fact, I would say many companies still use them and are a huge part of their technology stack. When x86 came strolling along and promised the ability to replace mainframes with cheaper and smaller hardware that required less overall investment companies became hooked! Fast forward to 2015 and we are seeing a similar change, some at the physical layer but more at the logical layer.
Back to pressure. I almost forgot that was part of this post since I have been ranting.
With so much new technology being deployed at such an aggressive rate it’s really hard to be an SME at all things. I can’t say I know everything about Ceph, there are a ton of moving parts. I can’t say I am an expert when it comes to OpenStack because there are multiple distributions and multiple projects within OpenStack itself. I cannot say I know every single specific detail on how software-defined networking works and what is best to implement as it depends on the infrastructure, use case, and hardware. At some point, you have to sit back and trust members of your team to be the subject matter experts.
These people are the ones you trust the most. They take the pressure off when you need to make difficult informed decisions. They are the ones that suggest solutions and how to implement such with the lowest amount of risk and the greatest return on investment. They are the experts and they should have your full trust and the best interests of the company in mind.
I admit I am a very technical individual especially at my current level, but I am by no means an expert of all this related to software as a service. If I was, I wouldn’t need a team and I could do it all myself. Need to program some stuff is Ruby, I got it. Need me to write some stuff in Python, no worries! Need me to write some bash stuff, simple as pie. Need me to develop Puppet modules to deploy your code and have intelligent rollback capability, a piece of cake. I hope you understand I am being somewhat sarcastic but if you can excel all of those things then you are amazing! Want a job? No, seriously… do you?
A healthy job is likely to be one where the pressures on management and employees are appropriate in relation to their abilities and resources, to the amount of control they have over their work and to the support they receive. I do not believe health is the absence of disease or infirmity but a positive state of complete physical, mental and social well-being. In a healthy working environment, there is not only an absence of harmful conditions but an abundance of health-promoting ones.
Work-related stress is usually caused by a poor organization (the way jobs and work systems are designed and how we manage them). For example, lack of control over work processes, poor management, and lack of support from colleagues and supervisors can all be contributors.
I for one am a workaholic. Yes, I admit it. I am in a 12 step program to try and get back to a normal life and detach myself from The Borg collective. It’s tough though with so many emerging and new technologies coming that I am excited about, but I am trying my best for my health, both physically and mentally. The technology isn’t going to disappear overnight so it’s better to learn to pace ourselves instead of trying to run a marathon as a sprint.
Find healthy ways to relieve the pressure. Have an open door policy with your staff, team, manager and other employees. I think one thing we overlook in our current era is talking. We instead hide behind chat, email, and other electronic forms of communication and are slowly forgetting to be human. I blame Google for all of this. Joking! As we enter the age of software as a service, let’s try and be more human. After all, I am sure some/most of us have seen Blade Runner so we know how the story goes.