Banks bid legacy tech farewell as they sail to the cloud – but now all that infrastructure is in hands of the big three Shifting financial services to the public cloud risks creating an over-reliance on the "dominant" service providers, banking heads told MPs yesterday during an inquiry into IT outages in the sector. Speaking to the influential Treasury Select Committee, Graham Bastin, head of operational resilience at Barclays, said the …
Bah, I prefer Lego or Playmobil.
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3291456
"They are going to store my money in the cloud"
Safety deposit boxes are going to still be real world.
Oh you mean your bank account? That's not your money, that's a debt the bank owes you. The ledger being on the cloud rather than a piece of vellum is probably safer, but if the bank folds, you're relying on deposit insurance.
It is indeed. But, instead of creating one gigantic Single Point Of Failure by putting confidential financial data into The Cloud (TM), they could have shored up their IT by not putting muppets at their heads and getting competent people to . . oh, I forgot, competent people cost money and the muppets are friends who have been promised good things.
Well, looking forward to the inevitable headline : AWS goes down, London is paralyzed.
AWS goes down
"AWS" goes down? All the AZs in all regions all at the same time? Hundreds of DCs simultaneously lose power?
Where do you think this compute load was running before? Why do you think cloud is more of a SPOF than running your own DC or two?
There is a load of shite talked about cloud, mostly by the "cloud! cloud! cloud!" pushers, but almost as much from the "cloud is just a huge SPOF" and "cloud just means other peoples computers" crowd.
"The Cloud" has the potential to provide substantially more resilience than any single company on a similar economic scale, you just need to design your services to leverage The Cloud properly.
And there in lies the problem. The "resilience" that was previously provided as a hidden technology service within a company is now in the hands of the business, and if they don't put all that nasty monitoring and load balancing and availability in place then things are going to break. Cloud can be cheap if you want a Minimum Viable Product. But you're going to pay (just like on premise) to have something that will weather the worst of storms.
But you're going to pay (just like on premise) to have something that will weather the worst of storms.
Plus some more because the cloud operators are running businesses and hence will be adding their margin.
What would be interesting is what cloud is doing to companies VAT returns. Previously, with internal IT, VAT would only be charged on purchases, not no staff etc. Outsource to cloud and VAT will be charged on everything.
Personally I'm not buying this talk of a "backup" away from the cloud. These are the same clowns that couldn't back up their own financials and they're banks. Besides, the sheer volume of data and transactions would mean a 1-1 live backup with the cloud removing any cost savings making it a cost and based on how long they have been running legacy systems I just can't see them doing it.
At least one IS doing it, with multiple independent private links supplied by different carriers between several of the big cloud providers.
No, it isn't cheap, but it is recognised as critical for the relatively noddy services they are considering deploying in a hosted environment. Core banking is not going to the cloud just yet however.
Cloud can mean one of two things. Firstly it just means running a workload on someone else's computer. Secondly, it could be using someone else's instance of a piece of software.
In both cases, using a cloud service doesn't mean that your service is now more secure that it was when it was on prem.
Cloud is not a universal cure-all. In the right circumstances, it has its place in your IT portfolio.
How do they expect to make it more resilient in the 'cloud' if they can't do it in their own data centre? Resilience in the cloud requires your software to be resilient, handling failure, so if it is, why didn't it work in their data centres?
If it isn't, how is moving out going to do anything, they will have to rewrite it, then if they are doing that, why move it to the 'cloud' they have fixed the reason for the move.
Bank IT was stable and just worked? You know the days of tightly coded, and tested jobs running on a m/f with terminal access, ah the joys.
But now we must migrate from "Legacy" and hire some Java and PHP numpties to create "glorious" and "modern" web frontends [and backends apparently].
Disclosure: I'm ex-Unisys, and also saw the horrors at another firm when they moved away from their IBM 3135
Weakest link of any cloud provider = their internet (copper/fiber/whatever) backbone going in and out of their data centre.
So... if you want to inconvenience a couple of companies, just pop the backbone (using a backhoe or something like that) and lots of people will run around screaming.
Greater success can be had by co-ordinating this kind of nefarious activity at other data centers.
Oops, I'm now on a watchlist, am I? >dons tinfoil hat<
Until bits get magically transported from one CPU to the next, however you provide your services is at risk of having their connectivity removed, whether that's cloud or on-prem.
If your service is mission critical, you should have it in multiple regions and availability zones - each which has multiple independent power, cooling and backhaul and be physically separated - and design your applications and data storage around not relying on any one region/AZ being up.
"Cloud" is more than running your programs on other people's computers.
I dont see how increasing the latency/transmission time between the application gui and the disk/cpu backend is really going to make a more reliable system.
Putting aside replication and the like, having a box sat in the banks DC room is no more unreliable than having the box sat in Amazons DC .In fact, probably less so as you have OpenReach and a man in digger doing road works.
Now the Cloud do allow you to hedge your man inthe digger a bit, allowing both increased capacity - more grunt - and increased geographic location hedging.
But that comes with with a price tag of something more C suite dont graps - you now have to write a distributed transnational system. See, a big IBM lump of iron may have been expensive, may have cost more per TFLOP than a lorry load of PCs ... but it works! Reliably!
Its not the hadrware, or where the hardware is located - ITS THE FUCKING SOFTWARE.
And, no, .Net does not really work. Yes, I know the nice man with PWC, who has a direct line to MS, said different thigns but he really is lying.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
