You don't need to be an 'investor' to invest in Singletrack: 6 days left: 95% of target - Find out more
Been doing an entry level role for six months. Pretty basic back up work with the odd bit of net back up and creating H files. I'll be going on training courses in next few months as I've no prior knowledge.
The main thing seems to be AIX here and MIGHT be some Oracle work later.
What are the prospects like in this area e.g. networks and servers? Good average salaries? Useful all over the UK and further afield?
Thanks
The good thing about *NIX skills is you can branch into several different areas relatively easily if you're of an analytical bent. So many things have them under the hood that it becomes relatively intuitive. You could even look into network engineering type stuff if you're so inclined. Worst case, you'll become a better server admin because you can speak to the network guys.
Prospects are good for server admins, but like any other area of IT, the more strings you add to your bow, the more desirable you'll be and the better your prospects for career advancement.
What @curious says, imho (as a 50+ year old) is that technology can and will change so it's about skill sets and always learning and adapting
Get yourself some Puppet/Chef/Ansible training and be prepared to be headhunted.
And actually get involved in these projects, contributing back improvements. Best way to improve your profile I can think of. Be part of the ecosystem rather than a consumer.
Rachel
@hagi
Any good online resources/books you can recommend for a novice please? For... errr... research. Not that I'm bored or anything.
[quote=allthegear ]And actually get involved in these projects, contributing back improvements. Best way to improve your profile I can think of. Be part of the ecosystem rather than a consumer.
Sadly, when discussing a role recently where they were interested in java and javascript, the only published work I could point to was the chrome browser extensions for this forum which I've contributed to 😳
I'm a Solaris admin. I was lucky early on as I worked for Sun, so was exposed to lots of nice tech straight out of Uni. Developed a very broad skill set which helps. A good SA should be able to turn their hand to anything...
There are jobs out there, but I'd recommend ensuring your Linux is up to scratch.
You're going to want to ensure you can shell script. People looking for linux/unix admins will also likely be after skills such as: VMWare/OVM, NetApp, git/puppet etc, SQL, Apache, Veritas Volume Manager, Nagios.
Next on my todo list is AWS and Docker.
I'd say average salary for an admin is £40-45k (central Scotland), you can obviously earn more if you then branch out into technical architecture/design, or go contracting.
The hardest thing I find is remembering stuff you did a few years ago when it crops up on a job spec, you know you can do it, but you have to convince interviewers. It's worthwhile having a lab at home to ensure you can freshen up if required.
Do you really need a lab these days with things like AWS and such around?
Thanks all. It's difficult to know whether to stick at this as a career. I'm from a project management non tech background so its not been easy at all.
[quote=curiousyellow ]Do you really need a lab these days with things like AWS and such around?
I just have Proxmox running on an old laptop providing VMs here (that's mainly because I'm adminning and developing a system based on that, but it would be suitable for setting up any sort of system to admin - I suppose I should try setting up a Solaris VM to freshen up my skills on that).
Following this thread as currently needing to resurrect a career, and sysadmin is something I'd be interested in. It's kind of on my CV - I don't have any of the specific technologies mentioned above, but I've done lots of installation and maintenance of Linux servers and Linux (various flavours) and Windows desktop systems, and quite experienced in Linux shell scripting. Any recommendations on particular technologies which would be easy and useful to learn for somebody with a C/C++ background?
Get yourself some Puppet/Chef/Ansible training and be prepared to be headhunted.
And Docker, vmware etc. There are, however, a lot less jobs for sysadmin roles (at least in Spain) than there are for developers.
Hi twixhunter.
I feel your pain as only in last 3 years I took a trainee Oracle DBA position coming from a problem management role.
So like you tons to learn including AIX,Solaris and now even SQL Server.
There's always new skills to have to learn in these roles so if that's something you enjoy or can get used to then that's half the battle.
Top tip - have a good strategy for learning. By this I really mean remembering. You'll do something new , move on to something else then have to do it again weeks or months later. I email myself code snippets and instructions and
Then move them into a different folder.
Good luck
As has been said, you need to get good at scripting as a UNIX admin. If you can get into the OVM side as well that's probably a good shout for the future (although personally I'm hoping Oracle screw themselves up so badly with their licensing bollocks that they go belly up, unfortunately I think that's wishful thinking).
Security is also where you can make yourself stand out from the herd, both from just general sysadmin work but also going into cyber security fields (there's a gazillion jobs going at the moment and a lot require a Linux/UNIX skillset). If you can end up getting security cleared (DV) then contracting becomes a lot easier (or you get good job security in a permanent role).
As for salary, £40-45k that someone quoted seems high - maybe for an experience admin but you'd be looking more at £30k for a junior admin role.
Well, seeing as there's a lack of resources on how to learn I'll put a couple up. There are the LPIC certifications which even if you don't get certified on will help you learn:
https://www.lpi.org/certification/get-certified-lpi/lpic-1-linux-server-professional/
It looks pretty basic, but it should get you going if you're an absolute beginner. Check out eBay/Amazon for the certification guides. There are also MOOCs on EdX on Linux admin which are helpful.
I think once you add some DevOps-ey type skills (e.g: Puppet, Python programming, some networking) then the world is your Oyster. Maybe low 6 figure sums in financial tech firms, and also gets you in the door with the tech giants (e.g: Google, Facebook) who pay quite comfortably in the 60s.
Depends on your motivation to be honest. If you're doing it for the money then you'll probably hate it eventually.
Loads of opportunity!
Not strictly UNIX, but check out the LAMP concepts, another vote here for Puppet. YOu can have a free version if using less than 5 nodes....i think....
Linux is and always has been my favourite OS and now its starting to come to the fore with companies sitting up and listening what it has to say for itself. Most cloud platforms are built on it and i can only see it getting more popular!
Theres a whole world of opportunity for it.
I'd say get some Puppet knowledge in and start claiming DevOps experience on your CV. It seems to be hot right now and good ones are hard to find.
Be good at it.
hey guys, our company is looking to move to chef/puppet.
We need to organise some training for ourselves. Any advice on which to go for.
I have the minutest bit of ruby experience. The other sys admin none.
I looked at some chef scripts etc.
The company is quite small and there is likely to a team of 3 linux sysadmins working on it.
I'd say we all a few years experience.
Im mainly a scientific Linux user and mainly manage the Grid Engine here and integration between the Dev Team and use of the Grid.
Do you really need a lab these days with things like AWS and such around?
That tells me you haven't migrated anything big to the magical cloud yet 🙂
Where I work migrated 12 servers running GPUs to AWS spread around the globe, first years bill was a bit over $800,000 dollars for CPU bandwidth, they are now back hosted in our DC's.
Unix prospects are very good if you can script and are happy to live in a world where nothing ever works as it should (I've been working in Unix/LAN/SANs since '97); I'm at the very high end of the Unix admin scale aka have a beard and no-one really has a clue what I do, or how I do it...
I've been working on migrating a lot of our kit from custom scripts to Puppet, which we chose over Chef due to our environment; I'm reasonably confident with Puppet (I'm writing our own providers and resources)
Only advice I have for either system is make sure you have a plan for how to layout your system before you start coding as its very easy to get yourself in knots.
If you are going to go with Puppet read the shitgary blogs, specifically on r10k, hiera (hugely important to enable code/data separation) and roles and profiles. Also read up on declarative programming and don't assume that the puppetlabs code is a. clever or b. consistent.
That tells me you haven't migrated anything big to the magical cloud yet
Not true! It's just that I feel like the days of having to purchase and run a shitty Dell chassis with 8 CPU cores or whatever while it took up space and power in your house just so you could fumble around on the GUI are done. You have a range of choices from cheap bare metal boxes for monthly rental, low end boxes and such if you just want to familiarise yourself with the interface. Learning a bit of networking would come in useful here.
Cloud services definitely have their place, just like owning physical infrastructure does.
I do understand the AWS model of "you pay for what you use" being a headache if you don't know what you're going to use. But there are cloud providers who let you "pay for what you rent" where you just pay a fixed cost for keeping your infra going if that is a concern.
Everyone I know who has migrated anything significant has moved it back out quicksmart within 18 months due to costs or random outages.
Its good for some stuff, but really really bad for others.
Everyone I know who has migrated anything significant has moved it back out quicksmart within 18 months due to costs or random outages.
yep, at work we migrated everything offsite after an IT consultant was paid a huge amount to optimise everything. Then after 18 months of nothing working properly, it has all been migrated back (cheaper and more reliable).
yep, at work we migrated everything offsite after an IT consultant was paid a huge amount to optimise everything. Then after 18 months of nothing working properly, it has all been migrated back (cheaper and more reliable).
Fits the Centralize -> De-centralize -> Centralize -> ..... cycle, shouldn't grumble though keeps me getting paid.
The sad thing was that our Linux sys Admin who ran everything kept saying 'you can't do that, it won't work' and was fired by the consultant for standing in the way of progress. Anyway, everything he'd said has come to pass and in the end he lost his job for just stating the truth / bleeding obvious. All because our then CEO loved very expensive Consultants....
The biggest challenge I keep seeing with cloud is migrating badly designed stuff (that works) into Azure or AWS without really changing the fundementals of how it works. The badly designed bits now stick out like a sore thumb where as before they were covered up by fast servers and storage or short network hops.
Meanwhile - yes learn to be a *nix admin, there is plenty of demand and most Windows only environments have plenty of Linux kicking about as well.
Pretty much, BigEaredBiker has it perfectly - if you migrate rubbish to “the cloud”\”somebody elses servers” then all the rubbish that hides quite well locally becomes really well exposed.
Sort out the system, don’t blame the hosting.
Rachel
@richc
Define significant?
Fully agree with "migrate rubbish out and it'll keep being rubbish"!
A bit disheartened about the sysadmin who lost their job for disagreeing with the a migration. That sounds absolutely nuts. Care to elaborate? Most people I know manage to disagree with company decisions without losing their jobs.
Either that, or he worked somewhere crap if people couldn't civilly disagree, in which case he's probably better off.
Care to elaborate?
The IT Consultant wore a suit and charged a £1000/day whereas the SysAdmin wore a T-shirt and was much cheaper (and ergo less wise).
Our CEO was never going to admit he was paying £k/day for crap advice, so just fired the sysadmin as he was the 'problem'.
Define significant?
Database servers, compute nodes, (very) large mailservers, webservers, large webapps.
Everything trivial or low load I've seen works really well, anything complex works for a bit then starts to go shonky. Blaming the application rather than the hosting is an interesting viewpoint, as cloud is supposed to save money which isn't going to happen if you have to reinvent everything pre-migration. Green field deployments are great, just not realistic for most companies that want to make cash.
Cloud + High CPU load = Bad news.
AWS and Database servers seems to be particular bad news, as from time to time your boxes get rebooted and then everything has to roll back to checkpoints and you risk losing data.
I don't care if people migrate to the cloud personally as its all just Unix admin to me, and the 'cloud' has caused a jobs and pay boom as ultimately this is just running on hosted hardware, which is my field ie: making stuff work as it should, and keep working as it should.
Just don't expect the cloud to be trouble free as its going to have just as many problems as running things in house just with zero transparency to what's actually broken when shit goes wrong, and zero comeback when you have an office full of people who can't work until somebody somewhere works out whats wrong.
Back on topic: If you are looking at Puppet you really need a development environment, and some decent revision control; so take a look at Eclipse + Gepetto + Yedit + Ruby DSL tools.
AWS and Database servers seems to be particular bad news, as from time to time your boxes get rebooted and then everything has to roll back to checkpoints and you risk losing data.
Interesting, this should be the same as on-premises - design the virtual infrastructure correctly and there should be zero data-loss (of committed transactions) even in the event of an unexpected node reboot.
The biggest 'mystery' challenge has been 'noisy neighbour syndrom' where nothing really changes on your workload but now eveything is running more slowly. I have two theories on how this happens;
1) Another company are on the same hardware/storage as you and just kicked off a massive data load/ETL process. This is what most people choose to believe.
2) AWS/Azure realised you are using their systems in such a way as you are an expensive customer, they live-migrate your VM's to lesser hardware or shared hardware with other 'bad' customers in an effort to make you upgrade your servers and become a 'good' customer. This is what most cynical admins believe 🙂
In either case this is where the good SysAdmins and DBA's prove their worth as they can quickly identify and tune application/database performance. If you work for a company that has an ethos of throwing tin at all performance issues the cloud will start getting very expensive!
Globally some of our dev/test work is done in AWS but no client hosting, Azure we do have some client stuff running (mostly as it was already there or part of the client's requirements when they came to us). Anything requiring additional security measures we wouldn't touch public cloud but we do have our own secure cloud and g-cloud offerings - that said it's too expensive to move our internal infrastructure onto it so we're not exactly eating our dog food...
We host web stuff on internal VMWare and run some (~500 boxes) on AWS - both have their positives/negatives. From experience shoehorning 'traditional' apps onto a bunch of EC2 instances, deploying your own flavour of a DB server etc isn't always going to give good results (to put it mildly). But when you begin decoupling things like MQ/object stores/search/mail and start leveraging managed services from the cloud provider, things make more sense. And if you're able to build event driven workflows - the 'loose coupled' approach AWS like to push - things get interesting (and cheaper). AWS Lambda is pretty damn cool.
Costs are a headache, and for us has meant skilling people up with the knowledge to run/understand relevant reports + crack the whip as needed. Can definitely work out cheaper though - a GPU cluster running 24/7 in EC2 will likely cost more than on-prem, but if you only need it for 8 hours a week, it'll cost less.
Anyhow OP - definitely good prospects. As said, get your head around Chef/Puppet/Ansible and maybe some monitoring stuff like logstash/kibana/splunk/new relic etc. From what we've seen more companies are making an attempt at DevOps, knowing the tools will set you in good stead.