With the numbers you are quoting on the backups, with the sizes of the databases, looks as though don't go through that much in changes. ~10GB of churn in a year is nothing for Sharepoint. I'm thinking you have a few numbers off. Ive seen it in clusters of 20-30 front ends with two massive database servers behind them churn 30GB a day.

Easiest way to get some scalability is to make sure your database server has enough horsepower. With that much data, a two proc/32GB machine should be sufficient. Get it into a Windows Failover Cluster, and you should have plenty of room to run.

The front ends are much easier to deal with. Load balance them using whatever and have them join the farm. Adding a web node is trivial in Sharepoint.

The rest is just knowing what you plan on doing in Sharepoint. If you are just presenting a website and using it for a CMS, nothing to it. If you are using any kind of analyitics to parse data, then there is more towards queries and database IOPS. If you are using it as a data repo, then giant database to hold the blobs would be important.

What are you running now? What seems to be bottlenecked?

@anonymous said:

There are not interested in a cloud option.

Why the hell not?

Putting it on any cloud provider would eliminate the whole ISP slow link issue, be centralized in a managed environment, and bring forth DFS pretty easy.

You are shooting yourself in the foot if you don't consider it.

As long as I am certain some folks are not there, then I'm good. Either I take the Charger for a run or I fly up, don't know.

Meh, then stick with what you got. Yeah, its more sexy to do it under any other SQL database, but if the end user experience is taken out of the equation, then f[moderated] it.

I've had to convert from a Lotus Notes DB to a SQL instance, so yeah, f[moderated] the users.

@BRRABill said:

I just need something simple, which is why I was thinking Access, maybe.

Should be able to covert from Paradox to an Access/SQL setup:

https://www.spectralcore.com/fullconvert/howto/paradox-to-sql

Biggest thing is setting up the user experience. Using Access run time would help a lot.

@scottalanmiller said:

"Pile" is not a term for a group of files

You just don't know the awesome power of the pile.

@KOOLER said:

@scottalanmiller said:

Anyone else seeing issues? We have some VMs down, Azure console isn't coming up.... doesn't seem like it could isolated to us or be account related if the Azure site itself isn't coming up.

It's interesting how Microsoft was pitching to own partners "now with Azure you can pack you things and go find better life somewhere else - we're taking away your SMB business from you"

People like to give shit to cloud providers, but this is the risk you take when you move off site. We had problems with Cogent back in the day:

https://gigaom.com/2008/10/30/cogent-sprint-un-peer-may-cause-web-slowdown/

Shit happens between two providers. You have to risk the positives of going to the "cloud" versus the whims of whatever your provider does. Far too many people think the interwebs is just magic and works. It requires lots of deals and such, like the Cogent and Sprint deal, which caused a lot of problems until it was resolve.

@scottalanmiller said:

@PSX_Defector said:

What you are talking about is outside of "Linux" and is done at the distro level. Hence why Ubuntu default installs garbage and DSL does not. RPM/apt/YaST, all of that is distro specific which controls that, it's not the Linux kernel doing it. And it's exactly the way that Windows handles it as well.

Right. Windows has one distro, Windows and it lacks RPM, APT and YAST style functionality. We are comparing the enterprise Windows release to enterprise Linux options (RHEL/CentOS, Suse and Ubuntu.)

And which we have SCCM or even WSUS or Window Update, which keeps up to date versus the whims of whatever repository where we are pointing to.

The reality of the world is that folks don't like to update. I had one customer who I had to cut over to a new server but used a beta version of PHP; Took a lot of convincing to change it up. Updates are good, but in the reality of SMB, you cannot find any way to update shit code.

@scottalanmiller said:

So... give me an example. Where do I run an install command that installs a trusted repo or similar, get's a major package that I need, installs third party dependencies in a trusted manner and then keeps them patches using tools from the OS?

Spiceworks. It installs what it needs, updates all the necessary components when necessary, and brings it up to date when necessary.

@scottalanmiller said:

@PSX_Defector said:

Because it's assumed that you already have the stuff available. Is it any different than running all kinds of updates to get a newer version of PHP, MySQL, or anything else on Linux? Yeah, you can script it, but that's about it. You still have to do all those steps on any *nix.

But you don't do that stuff on Linux, you see. That's what we've been pointing out. It's all handled by the OS itself, both the install, determining it as dependencies and handling versioning and patching. What you are describing is actually the misconceptions that Windows admins seem to always have - that all of this work that you are doing on Windows has to be done on Linux.

So when I compile from source a kernel install, it's just magically gonna install, configure, secure and harden third party applications for me? Yeah, I don't think so.

What you are talking about is outside of "Linux" and is done at the distro level. Hence why Ubuntu default installs garbage and DSL does not. RPM/apt/YaST, all of that is distro specific which controls that, it's not the Linux kernel doing it. And it's exactly the way that Windows handles it as well.

Just because your binary package management tool takes care of all of that doesn't mean that it's not being done. Dependencies and various other stuff for third party applications are perfectly built into the MSI installer process. It knows how to check and what to get, as long as the installer is setup to do it. Just because some folks don't make that happen within their installers doesn't mean that Linux is superior, because I've known plenty of RPMs that are not worth a damn in repositories. As we see here, crap installers are crap installers.

Run "netstat -ano" and find what connections are going out, what the system is listening on, and what destinations are coming in.

If you want deeper info, Wireshark that shit.

@scottalanmiller said:

I think that it is often overlooked that "hard" Linux installs often involve doing a huge amount of work that in Windows is views as an unrelated task (downloading and installing the platform, database, etc.)

Because it's assumed that you already have the stuff available. Is it any different than running all kinds of updates to get a newer version of PHP, MySQL, or anything else on Linux? Yeah, you can script it, but that's about it. You still have to do all those steps on any *nix.

When I deploy something other than standard .NET code on a Windows install, I make sure I have the right binaries in my library ready to go. And this is definitely outside of .NET. We don't even support the use of PHP on IIS. It's always a bad idea to use it.

Linux people think Windows is so complicated and hard when it's only them making it out to be. Then they deride us Windows admins because we only "click" and don't know anything.

@scottalanmiller said:

If you were doing this "this Linux way" you just "yum install snipeit" and done.

Real men compile from source.

Youtube Video

@scottalanmiller said:

They do everything. They staff ALL of CitiGroup IT now, for example.

Funny, I just got a call for a job at Citi. Direct hire too via a recruiter. Very strange.

At least @Lakshmana isn't going to work for Tata. Those guys really don't know what they are doing. They are the ones who when we notified them of their Terminal Server spewing Facebook spam decided the appropriate action was to put facebook.com into the host file as 0.0.0.0.

WiPro isn't much better, but it's certainly a good move up from the fly by night operations over in that part of the world.

http://www.theregister.co.uk/2015/11/20/level3_outages_reported_in_us/

Hit us too.

@coliver said:

@Dashrender said:

@gjacobse said:

Why wait till 2017 to move to O365? With DirSync and other tools, you can have SSO and all the joys of on premise, and not have to worry about a physical device.

Because I am already paying a three year SA Open License agreement for my current setup that expires in 2017.

Also, before I can even consider moving, I have to either move most if not all users to OWA or fully implement a complete replacement system for our current calendaring system for physician calendars that will sync with iPhone/iPad/Android.

Outlook cached mode is completely unusable for us. I quickly discovered when we converted from Domino to Exchange that cached mode wasn't anywhere near real time when updating other people's calendars. We require realtime (or very very close to) updates so that all 80+ people viewing calendars are aware of changes/updates, etc. OWA or a third party would work for this, but O365 with Outlook in non cached mode I would only image would be a horrible experience.

Have you tried Sharepoint Online shared calendars. I'm fairly certain they can readily sync with an Android/iOS device. Maybe not as a primary calendar but then again your third party app wouldn't be primary either.

That would have been my recommendation as well.

Exchange is designed around a set of users. Sharepoint is designed to share, hence it's name. And you can integrate them together

http://blogs.technet.com/b/ptsblog/archive/2011/05/31/sharepoint-and-exchange-calendar-together.aspx

Totally depends on what kind of backup scheme is needed.

Had a 42U tape library in each of the DCs back at the big red V. The system had 8 drives and the ability to hold well over 2 months of tapes. We still had bottlenecks with regards to streaming data to them. But we were a multi-tenant environment, backing up hundreds of customers.

For these small shops, tape makes sense for compliance issues. To keep backups offsite easily and with less of a chance for catastrophe. A nice D2D2T system will cover most everyone's needs.

@Dashrender said:

Joe's isn't a cheap place. The last time I was there (and it will remain the last time) it was expensive. I think the average plate was $28 plus tax and tip.

That's downright cheap, that's Darden cheap, and those are some cheap mother***[moderated].. Come back when entrees are $40. Some of the Brazillian Steakhouses, even the cheap ones, start at $50 a plate for dinner around here.

As for turnaround, on a 6 hour shift, at a decent place, you expect about two to three services per night on a good night, so assuming a full house you have three different sets of people at the table. Then it's just how many you get in your section, too many and you are screwed, too few and it's not even worth coming in. Lord help you if you get fundies coming in, tipping you with Chick Tracts.

@scottalanmiller said:

You can build a lot of drivers and such into a base image. Although standardizing on a model is a good first step. You want to reduce "Model Sprawl" whenever possible.

I haven't made model specific images since the early 2000's when Win98 and Ghost ruled the land. Since about 2005 or so I've stuck with creating a universal image in Vmware and deploying it to machines.

And since the consolidation of all the hardware with AMD and Intel it's been much easier to keep OOBE in line on boot. Used to be lots of different drivers would need to be deployed, but now I can just lay down basic Windows drivers, put the updated ones on a directory, then deploy it out.