In 2004 having taken the reins of IT duties for Windy City Parrot, I moved the website from a desktop server under Catherine’s desk to a server somewhere that I forget, but within a few months enough web hosts pissed me off until I ended up on a Wolfpaw server in Tenafly, New Jersey.
We’ve been in the cloud for 15 years.
In 2005 when our Milwaukee Street (Chicago) facility burned to the ground.
Had our website been on Catherine’s desktop server with all the local backup we would have lost the entire website and would’ve had to start from scratch.
With the site on a remote server we just needed new computers so we could talk to the Internet once again.
We ended up moving away from Wolfpaw hosting because of Miva Merchant, our e-commerce platform at the time changed its business model which would have tripled our monthly hosting expense.
The cloud is much different today because back then Google, AWS and Microsoft were offering cheap processing on giant servers that can scale up on a moments notice.
This not only helps large corporations like Netflix but small potatoes like us because cutting-edge technology has been made affordable.
Below you find an email thread between me and Jeff Koch who owns Wolfpaw hosting.
8:23 AM (13 hours ago)
to Mitch, Catherine
How the internet works – video
On 8/25/2019 5:13 PM, Mitch Rezman wrote:
On Sun, Aug 25, 2019, at 4:12 PM Jeff Koch <firstname.lastname@example.org> wrote:
OK – we’ll start the backups tomorrow. We normally set the TTL on the DNS zone to 10 minutes during these migrations so the new IP address will be effective very quickly.
I’ll keep you posted.
Wolfpaw Hosting LLC
On 8/25/2019 3:13 PM, Mitch Rezman wrote:
Great explanation – thank you!
Let’s do Wed, Our big marketing email deploys on Sunday mornings giving us enough time to fix anything beforehand.
No need to block as it may take time for the block to be lifted by the internet.
Windy City Parrot where the sun never sets empire 46356
On Fri, Aug 23, 2019, at 5:18 PM Jeff Koch <email@example.com> wrote:
After last February’s debacle, I vowed ‘Never Again’ and so far it’s worked out OK.
Fiber optic junction box (above)
(February 2019 workers cut the fiber optic uplink to Wolfpaw shutting all the servers down for almost 72 hours).
Yes, I think Amazon had a recent outage and Netflix was down for half a day. After that, the Google Cloud was down two months ago for four hours because some engineer entered the wrong code – and caused YouTube, Google Doc’s, GMail, Google Search, Slack and others to be off-line. But somehow I feel better knowing it’s not just me when that happens.
Both GCP and AWS create server instances in the cloud that simulate dedicated servers that are all fully isolated from each other. The resources are fully backed up with RAID arrays for disks and similar for CPU’s and RAM.
Everything is monitored so that when their systems sense that disk, ram or CPU is failing you are automatically (GCP is fully automatic, AWS you get an email asking you the restart the instance) shifted over to new hardware and the instance keeps running.
So you could say it’s like a virtual dedicated server but better in the sense that it is assembled from resources that exist throughout a massive data center rather segments of one server.
Editors note Windy City Parrot uses G Suite for our email and other uses.
On 8/23/2019 4:52 PM, Mitch Rezman wrote:
Sure, we both know we should be in need to be in the cloud and what could possibly go wrong?
At least we know no more uplink breaks.
Will this be a scaleable VPS now or still dedicated?
Do I need to redo my SPF/DKIM/DMARC settings?
Although available (and expensive) In my mind there’s no point in getting a dedicated server because (a) you lose some of the advantages of the above and (b) with the above if you need more power you just dial it in.
We still take daily and weekly backups that are stored on multi-datacenter disk arrays.
Your mail is handled by Gmail already and I don’t see anything in your SPF record that would need to change although I just updated your SPF record to allow mail from the website. DKIM is not affected by the location of the websites.
You don’t have a DMARC record but if you want one I’ll put it in.
Your old box will remain online for maybe a week after the move and then it will be in storage in our office. Not sure what to with all the physical hardware but I’ve got a boatload of servers and several 100K invested in that stuff.
I’ve been running down to data centers in the middle of the night for almost twenty years – not sure how I’m going to live without that fun.
Anyway, you’ve got some big websites and I’d say we plan the migration for next Friday evening. You’ll know later that night or Saturday morning if there’s a problem and we can fire up the old server, if necessary. If you like, we can block access to only your IP address until you’ve checked it out – that will make it easier to revert if we have to. We can also do this Wednesday or Thursday evening if those are lower sales volume nights.
Let me know which days work best for you.
Wolfpaw Hosting LLC
I would advocate my old box not go away until we are sure Chu’s Zen Cart/Wordpress Frankenstein survives.
My corollary to Murphy’s law: Mr. Murphy was highly optimistic.
Fun fact: Netflix is hosted on AWS
Windy City Parrot where the sun never sets empire Tel: 312-492-9673
Mobile: (219) 401-5505
On Fri, Aug 23, 2019, at 12:53 PM Jeff Koch <firstname.lastname@example.org> wrote:
We’re ready to migrate your server to AWS.
You have about 150 GB to transfer between the store and the blog.
The plan would be to a full backup while the store is running, upload that to the server – that would take about a day.
Then a day or two later in the evening stop the server (or put the sites into maintenance mode), do an incremental backup of the websites picking up anything new or changed and upload that to the new server along with the database dumps.
Then after loading all the backups we change IP addresses in the DNS and fire up the new server and have you check things out.
Obviously there’s more to it under the hood but we’d like to get started.
Hi Mitch – we are on schedule to do the migration tonight. The sites will be off-line for 60 – 90 minutes while we take final incremental backups of the websites and backup the database, upload these backups, expand all backups and finally put the files into place.nags, be done in time for you to still check things out.
Wolfpaw Hosting LLC
I thought we’d start at around 9 pm EDT and, barring any snags
One hour after the migration completed:
Hey Jeff – The site can’t take money because it can’t resolve to the host!
There was an improper DNS setting
Jeff responded promptly after correcting the setting
I love moments of terror like this – glad I could fix it fast – but we could have always reverted. 🙂
Best regards, Jeff
On 9/4/2019 3:28 PM, Mitch Rezman wrote:
We went down for an hour Sunday nite but I think it was a hiccup – so yes retire the old server – thanx
4:37 PM (1 hour ago)to Mitch
You had two issues recently.
First, Sunday at around noon you started getting attacked from multiple IP addresses owned by the Huawei Cloud service in Hong Kong.
We ended up blocking those IP’s.
Second, Monday night the apache web server program did not restart properly while working on the Webalizer statistics.
We put some extra code into that program which should cause apache to re-try the start.
With respect to the first issue, we’ve had good success using the Cloudflare CDN that would handle DDOS (Distributed Denial of Service) attacks mitigating service for websites that get attacked frequently.
Basically you sign up for their $20/month plan and they act as an intermediary between the website and the world.
They block attacks while at the same time allowing good traffic through.
Cloudflare is a CDN
On Wed, Sep 4, 2019, at 2:20 PM Jeff Koch <email@example.com> wrote:
The new server seems to be working OK. Any problem if we take down the old server?
BTW I looked at Cloudflare
Their $20/mo plan is not effective for our 5000 + pages
In that WordPress (our traffic magnet) is a subdomain they require a complicated matrix of Anames and very specific security protocols.
I tried StackPath CDN, they crashed the site 5 times in 12 hours trying to get everything to work
I’m going to build a new site that requires little to no coding.
I know, “good luck with that”.
Thursday morning after the migration, our “Store Manager” would not connect to the site – here’s how that was fixed.
Catherine says the store manager now works.
We have the firewall at the server opened for access to the database access only for IP address 126.96.36.199.
Please find out if this is a static IP address because otherwise it will change periodically and access to the database will be cut off.
We would prefer not to open access to your database to the world – that would be dangerous to your website – and a vector by which hackers could create problems.
The other alternative is to sign up for a VPN service with a static IP. That would allow you to tunnel all your internet activity (office, laptop, home) through the VPN and show up as using one of their static IP addresses.
VPN’s are also good for security since they create an encrypted connection from your PC through the tunnel.
We’ll just leave it to work as it is