Do the remote mirror sites have to phone back to the mother ship in the bay area? Or can they operate completely independently? I have a lot of home automation, switches, lights, garage door, etc. If they can't reach the internet and phone home, they are DOA.

Ive been thinking about the transition to the cloud and disaster recovery.

Most everything I run today is either completely in the cloud, bits and data, or some hybrid.

For example, Salesforce, Azure and AWS. Without those, my company cannot operate. Many companies are in the same boat.

There are two major faults in the bay area: Hayward and San Andreas. They are due for a large shift. If that happens and the bay area is leveled, are we crippled to the point where we can't produce basic goods?

Yes, this is technical, but it's also political because I'm wondering if any politician is asking the Q and if not, why not?

We did a major ERP rollout a few years ago. 34 states.

We had 4 datacenters because of the issues you highlight.

Each of the datacenters could service all locations if SHTF.

TAMUallen said:

---

Your post screams crazy.

Don't put anything in the cloud that means anything to you. End of discussion

---

Aaaand, quit using Salesforce. Product is sheyit.

People have thought about this fairly extensively. Go read up on availability zones and regions. Not all systems and applications are built with the resiliency that comes with multi-AZ and multi-region, but many are.

I have a data center and across the street another….high availability. Suck it aws!

Any architecture like that should have disaster recovery plans in place. For companies that specialize in data centers, disaster recovery plans should be a key part of their business. If any data center shuts down for whatever reason it should fail over to another data center in another part of the country.

It seems like a greater threat would be hackers who want to steal your data rather than shut you down. Instead of hacking dozens of corporate networks one by one they can simply hack a single service like AWS and get access to everything at once. However it should be much harder to hack a large entity like AWS than a security layer supported by a company that doesn't specialize in secure networks.

If the Big One did hit California it would have major impacts around the world for the short and long term regardless of anyone's reliance on the cloud.

One other major concern about moving everything to the cloud is that your business is now beholden to the cloud service provider. It isn't quick or easy to move your entire infrastructure to the cloud nor to move it back on-prem. If the cloud service provider should take exception to something about your company, such as you didn't hire enough transgenders or you engage in wrongthink, they can shut you down immediately like they did with Gab.

This whole conversation reminds me why bitcoin is completely worthless. Suck it bitcoin bros.

When you're designing a disaster recovery plan for any system, these kinds of questions arise. Redundancy at every level is considered.

-Making sure your servers have redundandt power supplies, and each is plugged into different circuits.
-Multiple servers, typically with some level of overhead protection. If N servers are required, you would run N+1 at an absolute minimum
-Servers are in different racks in case one is brought down for any reason at all
-Servers in multiple buildings in case one building is unavailable
-Power coming into the data centers from different sources, different power stations, so that if one part of the grid goes out you still have power. In particular, the placement of these should be "on opposites sides of the fault line", in common parlance
-UPS and backup power for every workload you're running
-Geographic separation of sites, the standard varies but we aim for 800 miles minimum separation.

All of this applies to networking as well

-Each server has at a minimum double the network connections and bandwidth for its load, so that if half goes down there is no issue
-Cables run to redundant top of rack switches. We typically run 4 cables, 2 to each ToR switch
-ToR switches run to redundant switches
-Redundant switches run to redundant routers
-Routers have a minimum of 2 circuits for internet connectivity, typically from two different service providers
-These circuits need to come in from different locations, and should nto share any single pathway. For example, if both your circuits cross a river over the same bridge, you need to get a different one, becaus that one bridge can take out your internet connectivity

Everything I've listed above is fairly standard, and has ben for many years. The level you actually implement depends on the cost vs the reward. One major advantage of using a cloud provider is that they handle the redundancy on a macro level for you. This is a great way for startups to get maximum protection without requiring a boat load of capital to do so.

None of the major cloud providers are at risk of being down from slicon valley taking a hit. There are a few services which are global by definition, such as BGP. This is what facebook screwed up recently that rendered their services unavailable for several hours, while they had to get physical access to repair it.

ETA- forgot about data storage. This has the highest levels of redundancy of anything, since this is the one things that cannot be replaced

Anyone remember when TexAgs crashed and everyone had to get new usernames?

"The Cloud" is just someone else's computer. For web-scale providers like AWS, Azure, GCP, that means multiple data centers across multiple states/countries. Each DC has multiple racks on standby, multiple (redundant/diverse) power providers and fiber connections.

If you lose your cloud hosted service it is very likely that it's because either you chose not to make it resilient or the SaaS provider did.

That's when I got mine. And I still use the temp password they sent me however long ago that was.

If your company is using Salesforce, you are already under water. I've used it in two separate jobs and is by far the most time wasting tool I've ever used. I bet your senior leadership says something along the lines of 'salesforce is the source of all truth'. All that means is you spend more time documenting your day on salesforce than actually selling. Stupid, overly redundant CRM tool.

Rant over

If a solar flare takes the mothership down and you figured out how to save a copy of your data elsewhere. Great, maybe you can trade it for canned food.

Sid Farkas said:

---

This whole conversation reminds me why bitcoin is completely worthless. Suck it bitcoin bros.

---

You have no idea how Bitcoin works.