* Posts by Robert Morgan

12 publicly visible posts • joined 28 Aug 2008

'No BS' web host Gandi lives up to half of its motto... Some customer data wiped out in storage server meltdown

Robert Morgan
Pint

Others have summed this up very well. My thoughts are - Gandi should have system level backups to get their VMs/systems back up and running again - irrespective of billing for it - those services generate revenue and being available/not being entirely lost is a big part of a hosts reputation. That alone is worth making sure system level/VM restores are available.

Customer based backups - i.e. give me file X from Y time and date is different and should be billable. Additionally, customers should backup their critical files to another provider/platform in the event they need to restore, because that's just sensible.

Either way, using ZFS arrays and relying on ZFS Snaps to restore from isn't the best strategy, ZFS is great until it goes wrong and then it goes really wrong - and that can be admin or hardware failures.

The 3-2-1 backup rule would have helped here. Three copies, two different devices/storage media types and at least one of those being offsite.

From a Gandi perspective, that might have looked like ZFS Snapshots (1), Snapshots mirrored to another independent array (2), that array rsync'd offsite to another site (3). With that, if one and two failed, they could have restored the (3) to another active array and got back up and running again.

Customers could have happily had a VM on Gandi with backup to S3/Backblaze with a lifecycle to delete old files and a copy to their homenas/dropbox/something else. That'd have given them the ability to get back up and running anywhere they could aim DNS records to.

In reality, large scale arrays for just storing large volumes of sync'd VMs to are cheap now, they don't need hugely fast disk i/o, the access pattern is sequential, so a low'ish cost box filled with 8/10TB drives gives a low cost of restore. A host "shouldn't" need it, but in times that this, spending that few K gives them another option. When getting back up and running, options are exactly what you want.

A final point would be, never make a backup plan, make a restore plan, work out what you need to restore to get backup and running, then protect the hell out of it!

Apple: Er, yes. Your iCloud stuff is now on Google's servers, too

Robert Morgan

I don't think this is a major one at all. My feel on it is they likely either use AWS/Azure/GCP for regional purposes - i.e. if you're in a far-reaching region, it's easier to use their reach rather than build local DCs. The alternative of course is that they're just using it as a form of backup or cold-store data archiving.

If it's heavily encrypted at rest and in transport and relevant controls around access are taken into account says keys only ever reside in a vault on Apples side, then their isn't an issue at all.

So. Should I upgrade to macOS High Sierra?

Robert Morgan

No Info on the External GPU Support?

I'm surprised nothing has been mentioned about the external GPU support added in High Sierra. You can now spend ~£300 on an external case with a 350w PSU, PCI-E Slot and a Thunderbolt 3 Adapter. You plonk a proper GPU in there (Nvidia or AMD/ATI) and suddenly you've proper graphics grunt on a mac again (something none of the current machines really have). It even works with bootcamp/windows and can be used with convertors so a TB1/TB2 Mac can still have an eGPU to increase graphics grunt without upgrading. Lots of the creative/video types are looking at this for the video encoding grunt.

Secret HPE letter tells sales team and partners to keep selling Arista 'confidently'

Robert Morgan

Sometimes it's worth looking at the why and the initial cause. It's a bitter pill to swallow from Cisco, but a bunch of the people leading Arista are ex-Cisco. In a few other cases, Cisco have typically left things to go bang - even after acquiring companies.

I'd expect the reason they're trying to litigate them out of existence is because they know the tech works and there aren't many working options at scale for Datacentre Networking at this kind of scale.

The sharks of AI will attack expensive and scarce workers faster than they eat drivers

Robert Morgan
Holmes

Here's the crux of the argument

Who's currently telling everyone the world will get taken over by AI? Who's saying be cautious about it? Largely, it's the biggest firms in the world. While I have a huge amount of respect for these companies and the work/people/knowledge they have, they have a cavernous gap when it comes to dealing with smaller companies. Want to Speak Directly to Microsoft? Google? Facebook? Amazon (at a reasonable level, not call-centre scripted fodder)? You'd better be from a relatively large company/service house, otherwise they're not interested. Part of the reason they're not interested is because there is a massive amount of education/conversation to have with a relatively uneducated customer (in industry lingo/technology). These people now are still jaw-dropped at the very concept of cloud computing, but still aren't really sure how to make use of it. Big Data is still a twinkle in the eye to most of them, it's just not something that's even on their radar.

Talking of which, how many people on here (which is a very business tech focused site) would know the ins and outs of AWS/Azure? How many people know how to use Azure DataLake/ML, How about AWS EMR/ML? Big Data isn't an easy subject, machine learning is even harder. There aren't vast numbers of people that really understand how this stuff ticks, and that's within our own industry.

Now, imagine trying to work in a normal organisation, a company that uses technology to drive business change. Imagine trying to pitch into the team the AI piece, of how it'll change the world and what not. When you get to grass-roots, most of the "You'll entirely replace a human" so it's cheaper stuff falls flat on it's arse. It's a technology investment that'll for sure, At best, if it goes well, will increase the productivity of said human.

At the moment, there are simply too many variables, even in the knowledge economy (i.e. being paid for what you know/skills/experience). When it comes to manual labour tasks/real-world environment scenarios, this stuff is even further away from being ready. Imagine Tescos delivering your shopping by robot. How does it navigate the basement steps? how does it figure out where the apartment is around the sidepath? What about un-even ground? Importantly in other industries, how does it charge? Battery technology simply cannot run a robot, nor will be able to in the immediate future?

I don't see AI going away, I welcome it and think it'll help humans work better, much like other software has. I think the projects that'll work will be the ones that augment humans, I think the ones that hit the wall and bankrupt companies will (mostly, it'll work in some limited use cases) be the full replacement models.

I see this a bit like the concept of Mutually Assured Destruction (https://en.wikipedia.org/wiki/Mutual_assured_destruction). If two businesses exist in the same market, if one has AI, it'll win, if both have AI, they'll likely compete/be close to each other. How do you improve the AI/run better processes? You need humans still. Also, you've paid Microsoft/Google/Amazon a truck load of money in the AI Arms race.

Of course, I could be completely wrong, and if we're all replaced, it begs the question of what do you do with all the proles that cannot afford to buy your products anymore but want to consume the earths natural resources? Enter Evil Genius played by Samuel Jackson from The Kingsmen? The 1% go for an End of World Party while the planet gets "cleansed"? :)

UK IT consultant subject to insane sex ban order mounts legal challenge

Robert Morgan

Isn't a fair chunk of this against human rights and what not? I'd have thought that this is totally against the ECHR - which is still enforced today.

If he's been to trial and been proven innocent, I'd think someone within North Yorks Police might have a personal link to this case, this almost seems to be going beyond the call of duty? Typically getting a officer to spend lots of time looking into a case, especially one that doesn't involve financial loss, seems quite tricky. The question must surely be why are they investing so much time and effort into making this mans life a misery?

Permies sitting pretty as fifth of contractors see rates cut

Robert Morgan
Pint

DevOps is a thing and it works... kind of

DevOps is a thing and is true and does exist... It however isn't a magic bullet for everything. It's not like extreme programming (i.e. sitting two devs together, or in this case sitting a dev and op together).

From a Dev Perspective, you have to be very used to using proper check-in/check-out of code repos while working on projects. From an Op Perspective, you have to get used to automating things and making things repeatable, you're normally aiming for more "small/short" outages rather than less "long/prolonged" outages.

Essentially, it works really well with applications that can scale across lots of servers (i.e. large scale web apps), or software that's provided to customers via a website/api and that ilk of stuff. In relatively large enterprises, internal services can also be built in this way.

Like anything in IT, you have to consider if and why you should do the DevOps on something.

Most people structure a bunch of "Ops" guys that can either use 1) hypervisors 2) public/private cloud 3) containers - and importantly, can also do some scripting/tying things together with Config Management tools like Puppet/Chef/Ansible/Others etc. The job of the Op is to then automate the arse off everything in sight. i.e. that server that does that thing? yep, make it built itself and get the code on it, and test stuff works, then join a pool. Oh, and if possible, let us specify if it's production/test/dev etc and put the appropriate environment variables/datasets on it.

The point of the Op is to understand the codebase/system at hand, and make delivery of that system to customers entirely repeatable and easy to deploy/build. That then changes how you deliver stuff to customers, you can literally kill an environment and build a new one immediately and quickly, and know you'll get the same thing each time. Essentially, your infrastructure starts turning into code. Basically, it improves everything.

Additionally, you also normally bolt Project Managers/Product Managers/Business People into the mix and discuss what's being done next/what's important. In the end, though you need to be more sociable, you get a bit more of a voice into that's stupid because of Y so how about doing it with X instead.

It does work, but it's not for everyone/everything, some things just aren't worth the effort, though as the software/systems to automate get easier/better understood, I think more things will come into this methodology, because it eats ITIL for breakfast - if done well... Otherwise it sucks

Got a pricey gaming desktop from PC World for Xmas? Check the graphics specs

Robert Morgan

I'm similar on this. I run a 980ti/i7 4790k on a Silverstone SFX-L 500W in SST-FTZ01B case. That happily runs inside a 500w PSU and it doesn't have any issues. I guess as others have pointed out, it's the ability for the said rails to deliver a stable supply.

I'd also hazard a guess that the 600w minimum is aimed at hobby builders rather than OEMs who have some level of design and build expertise. In my rig, as it literally has the motherboard/cpu/ram/gpu/ssd, there isn't a whole lot of other stuff to power. I'd guess the 600w minimum is aimed as a safe bet with some headroom for a few extra cards/accessories etc.

As others have said though, I'd take a quality PSU over the cheap "750w honest guv" PSUs anyday.

I also don't fully despise PC World. They have a time and a place, that place might be minimal for us who are professionals, but it's still occasionally got us out of a pickle when we've needed something immediately.

AWS CloudFront wobbles at worst possible time

Robert Morgan
Holmes

Re: This confirms my theory

I'd not think that holds water to be honest. AWS has been in the making since ~2002/2003 onwards, sure, initially it was a view to helping run Amazon.com, but the ultimate aim was always to make it fully accessible and multi-tenant. They're killing most, and it's near impossible to keep a normal/traditional platform online nearly as much as you can with something in AWS. I think their capacity is well past the running amazon.com stage now, I'd cowardly hazard a guess, that amazon.com (and all the international sites) are < 10% of the compute load on AWS.

Seize your moment, Microsoft: iPad is RUBBISH for enterprise

Robert Morgan
Childcatcher

Consume not Create

The article is obvious, but it points out things that a few people are missing. Quentin hit the nail on the head. The iPad - and tablet devices are FANTASTIC for Consuming content, but they royally suck at creating it. The main reason for that is we've been using point and click mouse interfaces for 20 years and moving to a touch interface is still totally unnatural.

It's fine for clicking a document and opening it, trying to modify a spreadsheet is a nightmare. It's all about requirements and being fit for purpose.

I'd like to see iPads work as a Multi-User device - but that's something Microsoft will do, Surface will plugin with AD and you'll essentially be able to have an army of directory enabled tablets. Then it'll become more interesting.

The whole point still boils down to the fact that it's a Consume, not a Create device though...

Boy band sings praises of Windows 7

Robert Morgan
Paris Hilton

It hurts

You seriously need to offer brain-cleansing facilities after exposing your readers to that kind of content! Maybe one for the advertisers to get in on?

Paris because her videos are slightly better.

McKinnon loses extradition fight

Robert Morgan
Paris Hilton

Dragging And Screaming

Well, at least he's off on the dragging and screaming option....

Tis total balls though, the guy was in the UK at the time - the crime might have been on machines on US soil but still, the law changed since he committed the crime. Its a bit unfair to drag him to the US.

Hacking into US Military stuff is serious but honestly, if its that important it should be on a private network, not the internet... live and learn, if the same happened to a US company from a hacker in the UK I doubt the same rules would apply here.

Paris, because Gary may be visiting a prison near her shortly.