@dashrender said in Where are MSP managed on-prem workloads moving?:

@pete-s said in Where are MSP managed on-prem workloads moving?:

Thanks, it does makes sense to move to SaaS solutions for a single customer that is doing their own IT.

But a MSP is in a different position because they, besides know-how, have a larger scale. So it can make economic sense to host things for their customers that doesn't make sense for each individual customer.

For instance does it makes sense for a company to have a server to host their website on? No, it doesn't. But if you're an MSP and your customers have a thousand websites that needs to live somewhere, it might make sense for you to host them.

I guess it also depends if you're an MSP that just manages things or if you also have your own hosting/cloud infrastructure or use another provider for that.

All good points. I have no view into that world, the few ITSPs I know are using other companies solutions, not rolling their own, or even hosting their own. Though some of them, we'll take JB for example, do manage all the stuffs other than hypervisor and hardware for things like a Ubiquiti controller, and PBXs.

If you really do have need to host 1000's of websites (or really massive sites, it could make sense to manage the whole stack, but then again, it could be better to get services from someone like Vultr, or in extreme cases like Amazon/Azure.

It's possible that ITSP/MSPs in the SMB space in general don't own any infrastructure themselves.

I know large companies that fully outsource their workloads to service providers. Those service providers host the workloads primarily in their own datacenters but also on public cloud infrastructure. But these service providers are often large companies themselves so they have scale.

@Pete-S said in Turns Out My New Client Has Colocation At CloudAtCost Datacenter In Waterloo:

@scottalanmiller said in Turns Out My New Client Has Colocation At CloudAtCost Datacenter In Waterloo:

@Pete-S said in Turns Out My New Client Has Colocation At CloudAtCost Datacenter In Waterloo:

But where are the proof for what you said?

Besides their corporate filings and postings online? We aren't claiming anything that they've not announced themselves. It's public info, not something in question or disputed.

Well, if it's truly the datacenter that is shady, then I would strongly urge my customer to move their gear elsewhere as quickly a possible. Locked rack or not.

Exactly, that's what we're thinking. It's the datacenter (aka Cloud@Cost) that is scary here.

Update: this is what I ended up with.
Route based VPN using this guide as a template.

Master site: 1x ER 12 + 1x ER 4
Sites A, B, C & D :1x ER4 each location
Colo: 1x ER4 & 1x pfSense (SM x10SDV-TLN4F+)

@scottalanmiller said in 10Gb/s Firewall Choice for Colocation:

@bnrstnr said in 10Gb/s Firewall Choice for Colocation:

Looks like the ER‑8‑XG could also be a good fit if you prefer the EdgeRouter series over the Unifi stuff. Also slightly less expensive, and better performance.

https://www.ui.com/edgemax/edgerouter-infinity/

And ordered... we should have it on Monday.

From the only vendor offering prime?

@JaredBusch said in Setup: EdgeRouter 4 + co-lo + infrastructure:

@scottalanmiller said in Setup: EdgeRouter 4 + co-lo + infrastructure:

@JaredBusch said in Setup: EdgeRouter 4 + co-lo + infrastructure:

@scottalanmiller said in Setup: EdgeRouter 4 + co-lo + infrastructure:

@Pete-S said in Setup: EdgeRouter 4 + co-lo + infrastructure:

@scottalanmiller said in Setup: EdgeRouter 4 + co-lo + infrastructure:

@Pete-S said in Setup: EdgeRouter 4 + co-lo + infrastructure:

Can't edgerouter do
failover?

As in a live/live cluster? That uses VRRP and yes, they can.

https://help.ubnt.com/hc/en-us/articles/204962174-EdgeMAX-Virtual-Router-Redundancy-Protocol-VRRP-

Then that is the way to go.

I would agree.

It is the best thing to do. Sadly I have never had the leisure to actually test it. I have a pair of ERL in colo but they were put in place long before this feature was added.

Yeah, but I have been authorized to make some upgrades that will free up an ERL for us, and we have an ERL in our colo. So maybe I'll ship one out there to do this soon. That would be an awesome project.

Don't think I would try and live test in colo.

085260da739d5f8723a626ab23a0da4623be9458998bfc91b38c57cdffec16d4.jpg

@stacksofplates said in Remote management of VMs hosted in colocation:

@dashrender said in Remote management of VMs hosted in colocation:

@stacksofplates said in Remote management of VMs hosted in colocation:

@scottalanmiller said in Remote management of VMs hosted in colocation:

@stacksofplates said in Remote management of VMs hosted in colocation:

@scottalanmiller said in Remote management of VMs hosted in colocation:

@eddiejennings said in Remote management of VMs hosted in colocation:

Allowing an SSH connection to the managementVM from the Internet

I have not tried this approach yet, and it appears more risky than the Screen Connect approach, since SSH to that VM would be open to the Internet. Unless I'm missing some benefit to this approach, I'll not be using it.

Use a strong key, lock to your IP. Very safe. Add Fail2Ban, of course.

Or add Salt and open/close based on need so it doesn't stay open.

Fail2ban doesn't work with keys.

But it would work normally with people attacking using non-keys, would it not? Or am I missing something about what it would do?

Why would you not require keys? Not making them mandatory defeats the purpose of using them.

I think he means - if a hacker is trying to use a password on a system setup to only allow keys - the fail2ban will block those users, or won't it?

No. It's dropped before fail2ban even sees it.

Oh, makes sense. There is no "attempt" like with a password, it is "already blocked."

@hobbit666 said in Monitoring low level server hardware health:

Might depend on the Server and what they have built in.

Dell have the iDRAC that you can log onto and look at hardware logs. (Not sure if you can setup E-mail alerts been a while since i've used one)

The new ones you can since version 7 I believe.

@dustinb3403 said in Small colo infrastructure for SaaS:

@pete-s said in Small colo infrastructure for SaaS:

This is how @dustinb3403 suggested replicating VMs between hosts:

0_1537834749485_colocation_network_vmreplication.png

While this is accurate, it also misses on the fact that he would still have his NLS server sitting, collecting backups on whatever schedule.

Other than that it is accurate. In Scott's proposal you are making the shift from migrating the entire workload (which is essentially instant) to migrating the database only.

In his case, the load balancer is the weak link in the chain. Granted these don't fail often but it isn't something you have control over either unless you provide your own for the COLO.

2 haproxy VMs (one per host) and keepalived for failover

@tim_g said in Networking and 1U Colocation:

fail2ban

Fail2ban does nothing with key based access. It's denied before fail2ban even sees it.

@nerdydad said in Redundant Network Connections:

We do have dual-wan connections, but we're also connecting 3 locations together. If we only had one connection connecting all 3 locations together, and one connection died, then that location (more than likely to be a plant) will be spending the day without any productivity because they wouldn't have access to the resources that they need. We're talking 250 people working at any one time at at least $11/hr.
11x250=$2750/hr (along with the added value they make to the company) versus $100-600/month for the extra WAN connection. It only makes sense for us to have the availability here to keep people working.

The City (another job I had) had every offsite pointing via Dish to Cityhall just to get internet. Some were 3-4 jumps. Single point of failure for the entire City. Every electrical storm took our 25% of our dishes.

Magic Unicorn Farts

@coliver said in Comparing Colocation and On Premises Infrastructure SAMIT Video:

First one of these I've listened to/watched. I kind of wish you would do an audio version of these with an RSS feed so my podcast app could download them. Shouldn't be too hard other then the hosting aspect of it.

I converted to MP3 you can get it from my Dropbox
https://www.dropbox.com/s/60qefvg632nnc5c/Comparing Colocation and On Premises Infrastructure.mp3?dl=0

@scottalanmiller

+1 to this, cause I once tried backing up ESXi VMDK to an external drive by plugging the external USB drive to the server...

hehe what a laugh that was, apparently after searching ESXi does not allow cases like this, cause it reserves all the USB and ports for passing through the virtual machines.

@ChrisL said in SMB, Hyperconvergence and Colocation:

How much do you think the "buzz" around the cloud plays into adoption from an SMB?

It's huge. So many places go to cloud computing and have no idea what they are buying.

1U and 2U commonly have "single server" prices. Once in a while you can get 3U and 4U prices but after 2U you normally go for a fractional rack. 3U+ servers aren't often useful for hosting. Not many are made and those that are typically aren't very dense.

@Dashrender said in Ease of the "Cloud", Without the "Cloud":

It surprises me that the costs could be lower. I would expect the economy of scale and lack of local IT support (or MSP/ITSP) requirements I would be allow prices to be driven lower. Of course you didn't say it would be lower, just said "often for much less cost."

Scale is what makes it viable at all. Elastic scalability is an insanely expensive feature to deliver. And neither solution requires a local MSP/ITSP, that's part of the comparison. These are "plug and play" level products. But in reality, using a service like Amazon where the scale is really good you run into a heavy technical barrier to use that HC does not have (normally.) So if anything, cloud makes you have more ITSP requirements compared to HC. Cloud has a lot of hidden costs that HC does not have.

Very nice, thank you.

@FATeknollogee said in ColocationAmerica:

@scottalanmiller How much rack space is being consumed, 1/2 cab, full cab?

Well, we HAVE a full cab. Over half is used on initial equipment delivery. We have plans for at least one large NAS to be added soon and possible one large Sparc server that we are interested in adding. There will be some empty space, but not a lot.

@MattSpeller said:

@scottalanmiller said:

@MattSpeller said:

@FATeknollogee said:

@scottalanmiller said:

The NTG Lab is prepping this week to a new colo, in fact!

In the Canada or NY area?

upstate NY, spitting distance to toronto

No, the nearest discussed location is near NYC. There are no datacenter colo facilities worth discussing in upstate NY.

whoops, figured it'd be at your hq

That's where it is now. Moving to colo soon.