Honestly the whole fabric of the internet, how email/SMTP and DNS and things work, is just a relic of an earlier time. I honestly think the money-men have their hands deep enough into the workings at this point that you wouldn't be able to create something like those things today and have them go anywhere. I'm surprised that it all still works as well as it does.
No, I was talking about the shared infrastructure. SMTP, DNS, ICANN, things like that require a level of cooperation and shared investment in the whole thing working well, not really because anyone's going to "win" the business game by running it to their particular advantage. That's a very alien way of thinking on the modern internet. The equivalent today would be something like massive publicly available caching web proxies that anyone could use as a big reverse-CDN to speed up their web access that were just kind of provided to everyone, government-funded, just sitting out there as a public resource. You know, like communism.
I've heard network engineers say they had a lot of trouble talking to their bosses about "peering" (setting up routes between two ISPs that happen to have operations close to each other, so they can hand traffic off to each other if it'd be more efficient to use the other guy's routes and both networks get more efficient to operate). They said they had a lot of trouble explaining the concept to the business people. They pay us for service? Fine. We pay them for service? Fine. We provide service to each other and both of us benefit without any money being involved? Plt... bzzt... I give up, I don't get it. Who gets paid? Why do we do this?
They've lost sight of the idea that it's a good thing to set up the world in a nice well working way (for everyone, including yourself), and just focused on how they can make their check bigger even if there's no point, or even if everything gets worse as a result.
Running personal active directory hybrid sync with azure, hybrid exchange, a separate red forest for management of vSphere infrastructure, using saltstack for Linux config management. ~50 VMs and containers.
This seems like a good use case for a cluster manager. I’ve used xCAT in the past and recently Lenovo has an interesting project called Confluent that includes a web interface. A paid option would be Bright. These are made to manage hundreds to thousands of nodes.
Look into getting a CMDB and keep track of all of your hardware. That can store the hostname / IPs of your KVM / OS or virtualization layer, vkernels, storage adapter IPs, your vCenters and so on and so forth. If your data is getting so big that spreadsheets are getting tough to manage, then you probable need a more enterprise method of storing it.
Check out NetBox. It is a free open source datacenter inventory management and IP address management tool. It will let you catalog all of your physical assets along with the network assignments.
Thinking out loud but wouldn’t chrome bookmarks for the URLs backed up to a file/account work better than a sheet of it’s just for access?
As we have mostly Windows based machines we look after everything is in Pulseway or TeamViewer. Routers and misc tend to be on specific ports on their connections IP and we have a shared Keeper repository for passwords and notes.
The company I work for has been buying other companies and customers like is silly season in the last year so we are digesting all the extra crud that came with it and trying to streamline half a dozen CRM, RM and Monitoring systems at the moment.
We use a separate subdomain. For example, all our hosts are joined to the ad.example.com domain, so remote management would be the same hostname on ilo.example.com.
We also have all HP hardware (at least for servers), so we have everything in OneView. Other devices (NetScaler SDX appliances, other stuff with management interfaces) just have their interface in that subdomain and it works out great.
At my company, we just have a standardized remote management suffix that we just throw at the end of the hostname, so we don’t actually track the urls. For example the server is named frosty, the url would be frostysuffix.
Then we track our servers with either an outdated access database that nobody updates, my locally saved personal Excel sheet, or by logging into one of the 4 different health checking applications that each monitor a piece of the infrastructure. (This part actually really sucks and I hate it.)
I manage 30 Esxi hosts with around 800 VMs currently on vSphere Enterprise licensing. Our company is preparing for the worst case by employing a 3yr plan involving:
Upgrading all perpetual lics still under contract to vsphere 8
(So we can run on unsupported vsphere 8 for up to 3yrs. if needed or until a resolution is found)
Assigning members from QC, Cyber security and Systems as an exploratory solutions planning group who report to the CIO and CTO.
(So we can explore different hybrid solutions, assign them for evaluation and give feedback based on those findings annually)
Hiring a Reseller partner of ours to do an audit plus an impact analysis in moving our environment from VMware to one of the exploratory solutions planning group recommendations.
(My company fancies getting ‘non-biased’ opinions from external sources, so we tolerate it)
Building active-active, multi -master, active-passive and active-failover hybrid solutions including those with SaaS vendors for our highest value systems.
(While expensive to do, this option gives us a clear nuclear level fuck you to VMware should pricing become too outrageous and we decide to pull out of renewal)
In the end, we will probably give VMware a 3yr probation period, regardless of cost and have a clear migratory path before that time should we decide that VMware’s TCO is no longer viable.
I was trying to get yacy working in a tiny container, but the dang thing kept crashing after indexing about 500,000 sites. Yacy is like a peer to peer web crawler. Too busy to dig into it and figure out why.
Sysadmin
Hot