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
Numbers do not lie.
It sounds like you are still in the free tier usage. Also I wasn't aware that they offer a managed service. Managed for instance like managed wordpress hosting. As far as I knew it was just like any other providers keep the infrastructure up.
If you are using lambda functions or other server-less functions it could potentially be cheaper .
Lambda it is.
You know that the hosting industry exists to keep content online 24/7 right? Having fast home internet will only change < 0.1% of the requirement. In any case, we've been on 10G for years, so once your home internet reaches that point, 10G should be pretty widespread with cheaper bandwidth.
Yes, cloud is cheaper at the entry level, but once you start actually utilizing resources, the price far exceeds VPS/dedicated. Typically we target customers that are pretty experienced in what they do and know that cloud will be $$$$$ for what they need.
That is not my experience and will depend on the use case. Also, managing a server takes valuable time and is at times stressful, which was my main motive to switch to cloud services. I didn't regret this a single moment.
So, you can just say a hired server, whatever the type, is always better. It just isn't.
Can't agree more. My employer spending $$$.$$$ on AWS which we can make it a lot more cheaper by moving from cloud to dedicated server but it wont happen, people love the scaling features, and proud to be part of the latest and new technology.
Huh? Broadband connections have been 24/7 for over 20 years. So then you're just talking about running a server/PC 24/7, which is already the case in nearly everyone I know personally and at work. And they're only getting made to run less power year over year.
Your argument is that higher bandwidth needs will be what's needed. And I ask, for what?
Having a single ISP and a single power feed is the difference. 99.9% of people use data centers because they need things online 24/7 and don't want to manage their infrastructure.
We have something similar for our own website
We are using Wordpress for backend, but our site is static site in PHP, does not fetches anything from SQL or Wordpress, our Wordpress installation is probably on another firewall protected server or may be some random directory with password protection or may be something else, won't say in public
We hired a developer to do this custom coding for us, when we save/update the page in Wordpress, it updates the static site accordingly, we are able to add new packages etc, update images, add new pages etc... and won't need a web developer for minor changes and don't need to manually edit html/php pages for such changes
Maybe today. Tomorrow, everyone has spare CPU capacity and reliable internet connections. Free CDN'S. My internet connection stays up in a hydro outage with a basic UPS for routers and servers. I'd much rather backup to my friends and colleagues where if I need to recover from fire or theft, I drive 10 minutes with a 10TB external drive.
The context of my post was for personal use (e.g. bloggers), not business where they need more than one server in more than one location with complex needs (infrastructure management). That's why I say datacenters are fine but hosting has less demand as people can reliably do it at home using existing hardware and services.
I am just curious to know how Comcast and Verizon will react to ddos attacks. They can barely maintain standard traffic flow.
Pretty much most Wordpress guest full HTML page caching plugins do this to varying extents but differing on whether PHP handles both the caching logic/cache serving and cache file creation. But probably not to what you are thinking.
One of these plugins I use is Cache Enabler https://github.com/keycdn/cache-enabler for Wordpress full HTML caching which has 2 modes
The latter more advanced mode is more performant when paired with Nginx and when you enable pre-compression of cached files and enable Nginx gzip_static and brotli_static to allow Nginx to serve the pre-compressed version which is faster than dynamic on the fly gzip/brotli compression for both request throughput and response time latency by a factor of 3x.
I wrote a guide on how Centmin Mod's Wordpress install can auto install Wordpress with Cache Enabler advanced 100% bypass PHP full HTML page caching at https://blog.centminmod.com/2020/09/06/203/wordpress-cache-enabler-advanced-full-page-caching-guide/
You can jump straight to the benchmark test results up to 1000 user concurrency test on $5/month KVM VPS server at https://blog.centminmod.com/2020/09/06/203/wordpress-cache-enabler-advanced-full-page-caching-guide/#7
Again, 0.01% of usage. IaaS is changing, not for the cheaper.
Great information, likely to attempt to implement on a test site soon.
With very high likelihood: yes, but if things really were even close to how your favorite youtube system-porn and entertainment sites make it look like you'd not be far off.
What has been the main driver for higher bandwidth need? Ever fatter, more bloated web sites and, to a far, far lesser degree more users/visitors, plus of course youtube, netflix and the like and to a degree faster wireless WANs (e.g. LTE).
On the other hand, how does available backbone bandwidth grow? Answer: usually very slowly plus it's expensive. Backbone bandwidth has grown dimensionally slower than both, last mile and "needs" (driven by ... see above).
WDWM evolution for example has a allowed for doubling both the number of channels and their capacity - but still, even the largest backbones are in the low Tb/s and it is very unlikely that that will increase dramatically (say to even "just" 100 Tb/s) within 10 years.
We quite often and reliably see considerable capacity increase in switching and routing chips and backplanes of ca. 12 Tb/s and even 25 Tb/s aren't rare anymore, BUT (big fat "but"): That's within the data centers which makes sense only to a degree as the backend can't keep up with the growth. Accordingly its main task is to make the "network computer" (as Sun understood and suggested decades ago, btw) faster that is, to make e.g. the front-end to DB, file etc backend faster.
What I see (in 10 years) is
So I guess what today's 256 MB mem, 1 vCore, 15 GB disk (SSD if you are lucky), 1 Gb/s shared $15/year VPS is will be something like 0.5 to 1 GB mem, 4 vCores, 25 GB SSD (or NAS based), 2.5 Gb/s shared in 10 years at about the same price.
On LET?
Does LET represent the hosting market? How many users here actually pay for aws, Google cloud, or azure? What percentage of content is delivered by those 3 providers? That should give you an idea of where the market is really headed.
Have not seen any mention of OpenStack - which is designed to do a lot of AWS type stuff. It is Open Source and well supported. Would think this could be used for the small Community Area Networks.
Once built it is very impressive....
The question is asked on LET, so was mostly answering to LET providers in mind. But it looks like the top 3 have 60% market share and it will only increase to a higher percentage with less than 10 big players overall.
So the smaller players will compete for the small fish. As costs increase for cloud hosting and companies and people experience better power, better Internet connections and lots of spare server capacity, the less need for using a datacenter.
Do you remember those IBM commercials where they'd replace a big ass room full of servers with like 2-3 cabinets? Just 5 years ago, a small server room that required AC and generator to run 8-12 Dell PowerEdge servers would be replaced by a single server that didn't require special AC and generator to run 4 hours.
It will always be in the big guy's interest to have the smaller guys locked into their ecosystems. Cloud hosting will offer more and more incentive to use them, until they've crushed the competition and no longer have to. Once your business is deeply coupled to AWS, for example, they can raise prices slowly and the pain of migrating out will be so high you'll just put up with it way longer than you should. So the future of hosting, for the big guys, is boiling frogs.
At the low end, we will see shifts to more and more effective utilization of resources. Container technology will improve to the point hosts can move back from KVM to LXC or whatever comes next, and clients will be happy with it. Clustered compute and storage will become ubiquitous, so cheaper consumer grade hardware can be used. Processors will have robust secure enclaves, allowing customers to have some protection from hosts sitting on their memory or disk, uwhich will be encrypted at all times. As the low end is more democratized, network performance and customer service will be the differentiators.
I don't see pricing changing much at all. The E3-1270 v6 server I started renting 4 years ago is still the same price today.
You just likely are just paying just enough to cover operating costs.
But is AWS really "managed"? In my experience, it's exactly the same as renting resources from any VPS provider, which means you are responsible for all software-related components of running a server.
I would consider "fully managed" to be the following:
I give a provider all of my site code and databases, along with documentation. They would install all relevant software, tune it, monitor it, and keep it up to date. In addition -- and this is critical -- they would put into place all orchestration (like Kubernetes) that's needed to scale resources up and down as needed.
Right now, this kind of service is "boutique" and costs major $$$$. Orchestration, monitoring, and tuning are very application-specific and require lots of invested labor. What we need is a bridge between IaaS, which gives you the resources, and the site or application to be run on it. If this became automated, easier, and cheaper to do in the future, then I would definitely consider moving away from unmanaged dedicated servers and VPS providers to such a managed service.
IMHO it is.
Server: provider is responsible for running the server, but you are responsible for running the software, including security, backups, etc.
Cloud services: provider is responsible for both.
I think cloud provider only responsible for maintaining the service. The software is at client end.
Can you give me an example of this?
I know there are hosted versions of databases and other specific services, but I don't see that the same as "taking care of the software" for an entire application or Web site. For example, it's not very hard to set up and maintain MariaDB if you already know how to work with it. I just don't see huge value in farming that out, especially at Amazon's prices.
If I rent an EC2 instance from AWS, I don't believe they have anything to do with updates or security patches. It's just another VM that I have to maintain.
Now, if AWS would take my Web site, analyze it, and roll out a custom, scalable, reliable hosting platform just for it and all of its associated services, then I would be interested. This is the hard stuff -- the glue that holds everything together. The heavy lifting is the full integration of everything, and all the automation that goes with it.
You are more talking about outsourcing instead of a managed service. AWS offers well defined services as managed like SES or S3. The outsourcing model is not scaleable like AWS or Auzue cloud is doing.
It is not about replacing servers (VPS, dedicated) by other servers (EC2 instances), but about using things like lambda functions, which are fully managed.
I am not sure if it is the best example, but see here for a large scale migration of a very well know organization:
https://www.bbc.co.uk/blogs/internet/entries/8673fe2a-e876-45fc-9a5f-203c049c9f9c
Possibly I am. But the subject of the thread was the future of hosting, and I see this type of fully managed service as a possible goal or next step. Rather than providing us with very well defined services or computing resources (building blocks), how about giving us a solution that puts those things together for us? They've already abstracted away things like storage and compute and database services, but can we bring that even further?
Essentially, I'm wondering if semi-automated orchestration is around the corner.
That might bring me to the AWS table.