Can someone explain to Wickes how the internet works

[zippy3x]

https://www.wickes.co.uk/

You have to queue to get on the website.

and yes, I've checked, my computer is 2 metres away from any other electronic device

[towser44]

zippy3x said:

https://www.wickes.co.uk/

You have to queue to get on the website.

and yes, I've checked, my computer is 2 metres away from any other electronic device

I think they are struggling. Their website was down completely yesterday whilst they decided what to do re the government position that they are an essential retailer.

I had a delivery from them on Monday of some paint and paint supplies and they forgot the paint! I was on hold to them on the phone for ages on Monday and finally got through but it was after 5pm by that time and therefore was too late to reschedule delivery for the next day. Of course, the lockdown then came in on Monday night. I was quite surprised my rescheduled paint delivery arrived this morning. The guy I spoke to on the phone was working from home, but had only just been set up and it was causing problems.

[Mr E]

Website under heavy load is managing access to ensure stability. Seems sensible to me.

I mean, my standard response would be “spin up more cloud resource”, but that may be unnecessary complexity almost all of the time.

Lots of DIY injuries coming up then.

[mondeoman]

Screwfix have pulled a huge amount of content - cant buy paint or door locks, which, given a 3 week WFH and good weather, seems a bit odd.

[zippy3x]

Mr E said:

Website under heavy load is managing access to ensure stability. Seems sensible to me.

I mean, my standard response would be “spin up more cloud resource”, but that may be unnecessary complexity almost all of the time.

Lots of DIY injuries coming up then.

Surely any website written in the last decade should be scalable. The precise point of the cloud is to spin up extra capacity when necessary. Like now.

As far as I can think, the only reason not to, is crap website design.

[Mr E]

zippy3x said:

Surely any website written in the last decade should be scalable. The precise point of the cloud is to spin up extra capacity when necessary. Like now.

As far as I can think, the only reason not to, is crap website design.

For sure.
However, they may have decided not to implement that (for whatever reason)
There may not be cloud resource available right now (for the price they’re willing to pay perhaps)

[Munter]

zippy3x said:

Mr E said:

Website under heavy load is managing access to ensure stability. Seems sensible to me.

I mean, my standard response would be “spin up more cloud resource”, but that may be unnecessary complexity almost all of the time.

Lots of DIY injuries coming up then.

Surely any website written in the last decade should be scalable. The precise point of the cloud is to spin up extra capacity when necessary. Like now.

As far as I can think, the only reason not to, is crap website design.

There's more to it than the front end website. Lot's more out the back end, some of which will have been put in place design wise, decades ago, and would cost huge amounts to change (years ago) for the ability to cope with an unprecedented event (now).

The demand management they have put in place seems sensible to me. Much better than just going non responsive or glitches happening. At least they are taking as much business as the system can cope with, rather than none.

[Stuart70]

zippy3x said:

Mr E said:

Website under heavy load is managing access to ensure stability. Seems sensible to me.

I mean, my standard response would be “spin up more cloud resource”, but that may be unnecessary complexity almost all of the time.

Lots of DIY injuries coming up then.

Surely any website written in the last decade should be scalable. The precise point of the cloud is to spin up extra capacity when necessary. Like now.

As far as I can think, the only reason not to, is crap website design.

Is unlimited cloud capacity free? No, didn’t think so.

[FourWheelDrift]

zippy3x said:

Surely any website written in the last decade should be scalable. The precise point of the cloud is to spin up extra capacity when necessary. Like now.

As far as I can think, the only reason not to, is crap website design.

They did it themselves.

[zippy3x]

Stuart70 said:

Is unlimited cloud capacity free? No, didn’t think so.

Literally every other retailer with an online presence thinks the cost of scaling up is returned in increased sales.

I didn't wake up this morning and think, I wonder if Wickes' website is working, I went to buy something. Now I got it elsewhere

Seriously, is that why you think they're queuing?

[CzechItOut]

That's nothing. Go onto the Ocado website right now:

"You are position 11852 of 11878. Your wait time will be more than four hours."

[Munter]

zippy3x said:

Stuart70 said:

Is unlimited cloud capacity free? No, didn’t think so.

Literally every other retailer with an online presence thinks the cost of scaling up is returned in increased sales.

I didn't wake up this morning and think, I wonder if Wickes' website is working, I went to buy something. Now I got it elsewhere

Seriously, is that why you think they did it?

Do you think nobody at Wickes IT thought about it?

Or did you perhaps think that if they are running at 100% of their capacity anyway, they'll be making profit on the things they are selling to the people using that capacity.

How would you feel if the website was up and letting you fill up a basket, and taking your cash. But even though the stock existed there were not enough people turning up to work (because ill/isolating) to get it either to the store, or to you. Would you be more, or less angry? Given you were quickly able to go to another retailer and buy it there.

I'd sooner a retailer say "sorry we're just too busy now, try a competitor". Rather than take my cash and let me down.

It's not as simple as just spin up some more web servers. There's the databases, stock control systems, personnel systems, 3rd party courier systems, bandwidth to the in house systems, IT staff off sick, then you get into the physical logistics of actually getting trucks in and out of places etc etc. The whole company is tuned to run at a maximum capacity of "x". You throw a whole bunch more of something randomly into that mix and it's pointless. You'd have to identify the bottleneck, and then assess if it can be improved, or not.

As an example Facebook found they were having to design the correct number of loading bays into their datacenters, in order to keep up with the constant in and out of, new/old hardware. If there were too few, they'd run out of IT capacity and have to build a whole new datacenter. And they are not even selling anything physical.

You can't just bung a loads more webservers at the problem, any more than you can just wind up the boost on an engine for a bit more for more power. It just doesn't work that way.

[Harpoon]

There's anecdotal evidence some cloud providers having capacity issues in some locations / zones (I think it was Azure UK South mentioned on report by The Register), so cloud capacity isn't unlimited.

It's also great spinning up extra website capacity and taking in loads of orders but what happens if you can't fulfil them in a reasonable time? Is the distribution centre down on staff due to illness / isolation / 2m apart rule? That must be having an impact on order throughput.

[number2]

CzechItOut said:

That's nothing. Go onto the Ocado website right now:

"You are position 11852 of 11878. Your wait time will be more than four hours."

.. but when you get there there aren't any delivery slots... I'm not sure of the point, they should limit traffic to the spots available so ask for area first.

[stevemcs]

I was looking at some loft board so i could tidy the garage up, however the next delivery slot is 3rd April, B&Q is 23rd of April, it looks like i'll be leaving it. Pretty soon there will be no point in having the website if you cannot get deliveries and as much as I want to use the click and collect service its not worth the risk.

[quinny100]

One of the customers I look after is a £150M bricks and mortar and online retailer.

After closing their stores, their online orders went through the roof. Despite a decent amount of automation in their modern warehouse - for certain lines the goods can be picked, packed and deposited in the truck trailer without a human touch - they can't get the stuff in and out fast enough whilst maintaining social distancing for the warehouse staff and have taken the decision to stop taking orders on their website for the time being.

It must be costing them a fortune - but they want to do the right thing.

There's every chance this may not actually be an IT problem and is just being used as a means to reduce demand. I suspect those willing to wait in a virtual queue may well be more likely to complete a purchase.

[SS2.]

I ordered a load of fencing from Wickes yesterday.

No issues at all with the website (seemed perfectly normal), although the soonest delivery slot was next Wednesday.

[zippy3x]

Munter said:

Do you think nobody at Wickes IT thought about it?

I don't know, but I've been writing complex software systems long enough to know that either, a) the dev team didn't think about it, or b) the management ignored it on cost grounds.

Munter said:

Or did you perhaps think that if they are running at 100% of their capacity anyway, they'll be making profit on the things they are selling to the people using that capacity.

Don't see why you'd queue people onto the website if that were the case. Waiting 5 minutes isn't going to see empty shelves replenished, or sick drivers getting well. It's obviously about website capacity.

Munter said:

How would you feel if the website was up and letting you fill up a basket, and taking your cash. But even though the stock existed there were not enough people turning up to work (because ill/isolating) to get it either to the store, or to you. Would you be more, or less angry? Given you were quickly able to go to another retailer and buy it there.

I'd sooner a retailer say "sorry we're just too busy now, try a competitor". Rather than take my cash and let me down.

Again, bad system design. Any transactional website worth the name will not take your money until they know they can deliver (more of this below)

Munter said:

It's not as simple as just spin up some more web servers. There's the databases, stock control systems, personnel systems, 3rd party courier systems, bandwidth to the in house systems, IT staff off sick, then you get into the physical logistics of actually getting trucks in and out of places etc etc. The whole company is tuned to run at a maximum capacity of "x". You throw a whole bunch more of something randomly into that mix and it's pointless. You'd have to identify the bottleneck, and then assess if it can be improved, or not.

I've been in IT long enough to recognise the things we tell managment to stop them asking questions.

For a start the browsing, selection and basket of any transactional website should be abstracted from the ordering and purchasing part.

The only things in your list that you'd need to keep your (browsing part) website running are the product database and stock checking information. The rest of your list are only relevant once an order has been placed.

A properly designed sessionless (if possible) website to allow users to browse products, should if designed correctly, be relatively easy to scale.

The order system could be completely separate, and one of the best examples is Amazon. When you place and order with Amazon, they don't accept your order, they accept your submission. That then runs through the rest of your list of backend/real life interfaces to check the order can be fulfilled. Only then does the user asynchronously receive an order confirmation (or presumably a suitable contrite declined order) by email.

Munter said:

As an example Facebook found they were having to design the correct number of loading bays into their datacenters, in order to keep up with the constant in and out of, new/old hardware. If there were too few, they'd run out of IT capacity and have to build a whole new datacenter. And they are not even selling anything physical.

You can't just bung a loads more webservers at the problem, any more than you can just wind up the boost on an engine for a bit more for more power. It just doesn't work that way.

Irrelevant, If there was genuinely a problem with the world's cloud capacity then you might have a point, (and if that is the case I will apologise), but i think you'd see far bigger issues than Wickes website, if it were true.

Finally, B&Q's website is fine, so yes, I stand by my thread title, Someone needs to explain to Wickes how the internet works.

[Munter]

zippy3x said:

Munter said:

Do you think nobody at Wickes IT thought about it?

I don't know, but I've been writing complex software systems long enough to know that either, a) the dev team didn't think about it, or b) the management ignored it on cost grounds.

And I've been doing capacity management on IT systems for 20 years. And that's long enough to know most devs don't understand the complexity of the the rest of the situation.

[zippy3x]

Munter said:

zippy3x said:

Munter said:

Do you think nobody at Wickes IT thought about it?

I don't know, but I've been writing complex software systems long enough to know that either, a) the dev team didn't think about it, or b) the management ignored it on cost grounds.

And I've been doing capacity management on IT systems for 20 years. And that's long enough to know most devs don't understand the complexity of the the rest of the situation.

And have any of the systems you've worked on had to implement queuing to get on their website?