New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Seeing that it's not so uncommon for even the big players to have problems, I'm not always really sure why so many companies go with providers like AWS or Azure. It's also ridiculously expensive.
99.98% uptime is actually not all that great, so I wouldn't pay "extra" for that. I guess I'd like to know in comparison to what? No SLA? Personally I consider SLAs to be largely marketing and relatively meaningless promises until those levels aren't met, which may result in some sort of, probably small, refund. That's swell I guess, but I don't base purchasing decisions on promises, but rather their track record.
The reason to use any of the larger providers probably comes down to one's preference for some aspect of the service they offer. Perhaps the more important promise , at least compared to many LE providers, is the relatively long term stability and consistency of the organization and service. For me, this is quite attractive. I've never had a big provider go out of business on me, get sold to some other provider, renumber my system IP addresses, move service from one city to another, and so on.
This broad and unqualified statement isn't so meaningful to me. Yes, they probably have a lot more "stuff" and keeping things running is probably a much more complicated than the average LE provider operation, but they are big. That is quite natural. I actually do appreciate those big providers who at least keep it relatively simple from my perspective. I think DigitalOcean is a stellar example of this.
It all sort of depends on what you want, what you need, what you prefer. There is no one answer that applies to everyone.
Yes! Absolutely. Depending on your use case.
1. Risk Management
It's mostly risk management. I believe someone touched up on it earlier but "Noone got fired for buying Cisco" is a thing. If a tech manager recommends AWS and then AWS has problems, well "AWS has a ton of expertise so I understand they could have problems." But if someone goes with JoshuaHosting (fake name... I hope) and it goes down then the IT manager could have problems.
2. Tooling
In the case of Microsoft Azure, they have better tooling of services. If you're looking for pure "hosting" then sure, but Microsoft Azure and GCP has some decent toolings available to work within their infrastructure. Azure Storage Explorer is fairly nifty for what it does and is much easier to work within their ecosystem than just getting a few block storage instances on Vultr or DigitalOcean.
3. Actual Support
I've been involved in a few contract negotiations with Microsoft Azure (North of six figures). The support, skillset, expertise, and just the amount of actual knowledge they're willing to put behind their product (and the ability to bring someone else in cross-company) is just amazing. Their added-on value on top of just "hosting" is basically the premium we pay for. We make assessments based on expertise we need, hosting locally, hosting with a bare-metal provider, vs going cloud. Cloud just makes the most sense (for us) each time around.
Why not a LEB Host
Most LEB hosts manage a good VM system, but most of the time you'll always have some neighbors that could cause problems. Hope is that by paying a bit more of a premium, you'll have more "dedicated" access to those resources. I've initially put our (early) resources on a very popular LEB host, but as our needs scaled it was significantly harder to grow with a LEB host, especially with the fact that AWS/Azure/GCP have instant scalability included. So I used a LEB host as a backup devbox/Proof of Concept box and then we had to give the actual contract to Microsoft Azure for the scalability and advanced tooling.
So...
At the end of the day, buy what you need. For most of the people LE* gathers, a hosting provider who advertises on LEB is good enough for you. But I will say we are 100% getting our money's worth out of using Microsoft Azure's platform. The amount of value we get for paying them (vs going with a LE* host) is insane.
Many companies waste money in far more egregious ways than "overspending" on a "premium" cloud service. I use AWS in a very limited capacity, and I sometimes think of how much fun it would be to have the ability to blow thousands every month on it. Lightsail is quite interesting as the pricing is more reasonable, and you get to dip into AWS's more primary services in a simpler environment.
But my God, the pricing, it's just not meant to be.
Thanks for the insightful reply. It seems that the reason why I don't understand Azure or AWS is that I'm just looking at them from a different perspective.
I work at the IT department of a company, and we mostly manage services for internal use. In this area cloud doesn't make much sense. One manager wants to move everything to the cloud, but it just sounds a bit ridiculous since we already have our own "datacenters" in place and can provide resources for internal use much cheaper than if we were using something like AWS. For stuff like build servers for developers, the cloud doesn't really make sense imo.
My job moved to Azure a few years ago, and my team was all pretty happy with how well everything integrated compared to their cobbled together bitbucket/jenkins. Later we needed to expand because we had too many builds and not enough agents, and it's definitely nice to just click and add more simultaneous agents instead of getting another on-prem server set up. We're also smaller, though, and we had a single 'datacenter' of a rack in our main office. We still have 3-4 servers in the rack of stuff that would be cost-prohibitive to put in Azure, but as we've expanded over the last year or two it's been a lot less work for us on the hardware side of things.
Honestly the best part is that you're saying that you have the knowledge in-house that the calculation worked to run everything in-house. That's awesome and that definitely shows that something like AWS/Azure/GCP isn't something you need.
Our clients aren't tech companies. However, they use tech as a vehicle to help support their intended goal. They don't have the time/expertise/funds to build out a full tech department dedicated to supporting these requirements. I mean one of our clients has already invested a significant amount (billions) of money into the core/fundamental product of their business. Outsourcing a few hundreds of thousands of dollars for Microsoft (a year) makes sense if we get access to their experts + server infrastructure support without having to hire full time employees, purchase bandwidth, find vendors to source hardware, design/build a datacenter, etc. Tools like Microsoft Azure Machine Learning Studio gives you access to a ton of Microsoft's Machine Learning toolkits (they dogfood this) while only asking for you to pay for the server power used. We just outsourced the need for an infrastructure team having to manage SLURM Workload Manager + VMs + data management/storage solution. This is just one tool out of many that they're in the process of building.
It's nuts. I mean bare-metal hosting is still going to be a thing and important, but the way I see it, AWS/Azure/GCP are all investing a tremendous amount of money and time into building these kinds of turnkey business-focused solutions/toolkits without needing system administrator level knowledge and still charging AWS/Azure/GCP bare-level costs.
DigitalOcean, Vultr, and Linode don't have the resources to compete with them on the feature list. So they're a good alternative on the pricing end, but anything more advanced and your ability to scale on those platforms will come to a screeching crawl as you work around their limitations. It's why rebuilding/starting at AWS/Azure/GCP from the beginning is so popular. Yeah for the base hardware/cost ratio, you're not getting the best deal. But the ability to grow quickly without having the server hardware/tech be an impedance is quite the attractive offer. Also, for a larger company, what AWS/Azure/GCP asks for is fairly cheap and reasonable.
I was expecting to write a shorter response as I didn't want to drone on. But eh... Sorry about that. Just super excited on seeing how Azure continues to grow in this. As you can imagine, I've been very much impressed with Azure (and AWSs') various infrastructure and data science teams I've had the fortune to interact & work with. Unless you can find someone who can bootstrap and build all these tools in a platform-neutral manner where we can link in with DO/Linode/Vultr/Cloudstack/etc. APIs, the way I see it AWS/Azure/GCP will continue to own most if not all of these businesses.
A thing I also forgot to mention that my company does projects in the defence industry. Strict regulations apply there and you can't use cloud for that even if you wanted to, it must be on premise or in a data center that meets the regulations. But that's a different discussion, it has nothing to do with pricing, scalability, service quality etc.
Oh then yeah I totally 100% get your point on that. Adding in different data security/policy constraints will definitely impact the actual design/usability of your system where that makes sense.
Thanks for the insight.
For decades to come. They are going to stay.
Thats managing the performance ratio.
I'd say that's a strong no honestly.