Discussion in 'Business & Enterprise Computing' started by elvis, Jul 1, 2008.
Install pi-hole and block social media.
Yeah, nah. I'm not having individual users switch accounts every time someone different wants to ring up a sale.
Whats the cutoff?
https://www.itnews.com.au/news/hacker-destroys-all-data-held-by-us-email-provider-519206 - that did the rounds a month ago.
Did Distribute IT ever come back after being owned 8 ways from Sunday?
Is "To big to fail" a legit control for the (admittedly, very small) risk?
Then tell them to stop doing per-packet. Per-flow ECMP has no issues unless bug or your deisng is retarded (e.g. load balancing across 2 different characteristic links but even then the jitter should be consistent on a per-flow basis) etc. - how can per-flow affect jitter etc as long as its hashed the same way in each direction?
OFC bugs etc. - theres a fascinating long post by a FB engineer on debugging massive ECMP hashing issues across their underlay, it turned out to be obscure silicon/algorithm issue and took them months to run down. But proves the 'theory' that per-flow, assuming its working correctly and hashing is not too uneven, works fine. Every massive hyper-scaler is doing L3 ECMP in the underlay, Cisco says do it in their CVDs, VMware says do it in the NSX validated design. Heck VMware even gets you to create two logical VTEPs per host simpy to get more entropy. Back in the early VXLAN days, the ability to vary the source port on the outer UDP packet was a friggin' feature because entropy. ECMP has been the basis for scalability/convergence for modern DC fabrics since forever, heck, its no different from TRILL-based L2 fabrics and the ECMP of ISIS routes in the underlay.
I'm talking about DC use-case. I can see plenty of reasons why you wouldn't want to in a SP / WAN scenario (guaranteeing CIRs and SLAs, traffic engineering and tunnels, simply less multiple paths etc.)
Smart move, wonder why they left it so late. We were taking bets on F5's demise (unlike Citrix they are a one trick point solution pony) but this may turn it around as it takes out one primary competing vector head-on (the scale-out software-based container guys though who knows if other options like HA-proxy etc. may start gaining more traction). Still doesn't really fix the cloud problem.
There's a cloud vendor for that:
I'm envious your friends/family are savvy enough to pgp emails.
Good question, in that linked case, I don't see it killing other businesses (1. it's email, 2. you have backups of your own right).
A Google/Azure/AWS shutting up shop, would a. kill a large proprotion of the internet, b. kill Microsoft/Amazon (Google still makes more money from ads than google cloud).
Does MS expect everyone to, go. cool, O365 was a nice experiment, I'll take X'000 copies of Office again thanks.
Azure/AWS AND their customers have a vested interest in running properly, it's almost their core business.
should facebook/twitter fail, you'd have a lot of angsty millenials, and some zero-value startups fail (those relying on social platforms to harvest data). No big loss
you'd be amazed how many people run businesses over the social network, it's not pocket change anymore.
There is far more to an f5 then ltm. Every other module ( that cost way more as well) is perfectly suitable to the cloud.
leveraging your whole business on a single point of failure is a sound business strategy.
You'd think that a large Australian telco would have the design expertise to properly deliver voice grade networks, well yes and no. Inside their own network they did great, debugging the NNI to NBN was a nightmare for me to get resolved :/
Isn't that BAU nowadays?
that's not the argument.
Disagree. Show me a restaurant or cafe with a dedicated website any more.
Yeah, it's quite incredible. Show even a medium business the average bill for a professionally "developed" website, and they'll just go "fuck it, put up a Facebook page and be done with it".
This is the new normal.
Nobody said it was smart. Only that it's common.
Indeed, most common things are utterly retarded.
People spruking social media setup these days and charging stupid money, the one that tried to engage me was upset when I wanted a tiered paytable for results, considering I have no online presence any customer aquistion would have been theirs, yet they still declined.
Yeah, that blows my mind too. I genuinely thought social media was the bottom of the intelligence barrel, but it turns out even that's too hard for some, so now there are entire businesses that help businesses get on social media.
"Stupidity" is, without a doubt, the world's only truly limitless resource.
too hard, or they dont (or dont want to) have time.
Its not all about stupid people
All their other functions are being commodotised as well (Remote access, WAF, proxy, the only niche left they seem to have a lock on is doing weird things with ADFS and O365 etc.) from my POV but this is not my area so happy to be proven wrong. Our load balancer application delivery guys say F5 is getting killed as well, we used to move them by the truckloads and we don't hear peep for the last 2 years. They're all frantically re-skilling on Azure, Zscaler etc...
re: cloud sure you can put in an F5 instance but aren't most people either auto-scaling linux instances or using AWS magically elastic/redundant PAYG product (WAF attached to CloudFront, ditto with GSLB, ELB etc.)?
sounds v painful.... I'm telling you from first-hand experience that most Australian telco is no smarter than most Australian enterprise, the nimble/smart ones run by techies excepted (i.e. the old internode/IINET, superloop, megaport, the old PIPE etc.)
If they're anything like the dozens of small business owners I know who have admitted flat out that it *is* too hard, then it's very likely actually too hard.
Entirely amazing how people are clever enough to sign up to Facebook and re-share utter shit, but not smart enough to start a business page and post something that will earn them a dollar.
This is where I don't agree - if a SMB is not in tech and most of their customers are on FB why not FB page?
And even if FB gets nuked how hard is it to run up a random page quickly to replace it? Why do it unless/until FB is nuked from orbit?
Fucking boomers, they ruined facebook. But boomers are people too. I geddit, people ruined facebook.