Hacker News Reader

Comment by funkyy | Hacker News Reader

If one of those giants could make a user friendly service with proper, clean documentation, then that would be bulls eye. Both AWS and Google are really bad at this and they assume every potential customer have quite good technological knowledge (IT college + 1-2 years work experience type of knowledge) or have all the time in the world to dig in in to countless instructions and specifications. With a first look I am not sure what type of server should I buy to set up a simple website - if such a basic information requires from me to do even simple research - I am out.

This is where DigitalOcean and Vultr are profiting the most - publishing tutorials, explaining everything and helping to go step by step through the setup with simple notes and comments.

View on HN

If you're just looking for a single server to let you run a website, AWS isn't really a good fit for you -- and they probably don't want your business. They (as you found out), aren't really setup to provide you the kind of support you need.

When you outgrow Digital Ocean and want to set up larger distributed infrastructure that spans datacenter and regions for high availability (and you hire the people that can run it), then AWS (or Google Compute Engine, Azure, etc) will be a better fit for you.

I am actually a startup and time is very precious for me. Within a day I managed to set 4 server infrastructure on Vultr - I have database server, 2 workers and 1 web facing server. While this is nothing compared to what companies use AWS for, I will be scaling very quickly. And best thing? I wont be scaling to AWS.

Lets look at Vultr, that is more complete option compared to DO in my opinion. You get click away following:

-backups

-ddos protection

-internal IP

-storage servers

-dedicated resources servers

With this I can scale up to rather large size. While this is not enough for some startups, for MVP and Bootstraping startups that's more than enough. And frankly once I will grow big enough to hire tech guys, I will just move everything either to dedicated servers or simply to Rackspace or Softlayer for premium cloud solution.

The whole point of my comment was - AWS, Google and Azure - all are loosing to the ones like DO and Vultr in smb market. And while it looks irrelevant now, what stops DO etc. to scale up and offer more services? I bet if not due to their enormous growth rate, they would be already doing this but for a year or two they must focus more on scaling rather than increasing number of products. Once they are ready AWS/Google will be forced to either buy DO for crazy money or rethink their documentation, offering, approach, business model etc.

I don't think AWS is chasing the Small Business market (but they probaby want the Medium Business market), they are looking for the enterprise market. I doubt AWS turns much, if any profit on a 4 server implementation when support costs are factored in. Instead of 100 4 server customers, they'd rather have a 400 server customer.

If I had a small static site that could be served by a few dedicated servers, I probably wouldn't use AWS myself. But, for me, the power of AWS is the elasticity - I can scale from 100 servers to over a 1000 servers and back down to 100 servers in one day - and those servers are distributed across 3 regions. And I have nearly a petabyte of data in S3. If I ran my app on dedicated servers, I'd need to pay for 1000 servers to handle my peak load, and I'd need a large petabyte-scale storage array (well several of them for redundancy).

I'm probably biased in thinking that my use case is more what AWS is seeking since they provide a good API and powerful tools to enable exactly this use case.

But that is what I am saying. If there is a startup, it builds infrastructure that is easy to scale. In a year they end up having 500 servers. Are they interesting to AWS? Sure. Are they interested to move their infrastructure from quite good setup that works to completely new provider? Not likely. Medium sized high growing startups cannot really afford moving to new infrastructure until they got proper funding.

I can scale Vultr from 100 servers to 1,000 in 1 day as well (most time this takes is to click things through, not sure if AWS have buy XXX servers at once but if they do, thats good then).

And once again, I am not saying about small websites/static sites. I am talking about startups. Companies that can grow XXXX% a month for first few months, companies that are very careful where they spend money and where they allocate resources.

Good luck scaling vultr beyond 30 servers. My company is using vultr for over 1 year and we were running about 40-50 servers during peak load. Here problems start to appear very often:

- We nearly failed our launch due to failure to increase server capacity during peak traffic, Vultr didn't have any more instances in that datacenter. This happened a lot of times and support always answered "We don't know when we will add more capacity".

- Periodic network partitions. In the last two months we experienced a lot of network partitions both on public and private network, some of them lasting for about 2-3 days, our fleet beeing totally partitioned in 2-3 distinct blocks. Even new instances started had private network down (100% packet loss).

- Technical support. We opened a few support tickets asking when private network will be fixed. Usually they respond that everything is fine and I have a configuration issue. I reply with ping/mtr logs and finally they agree there is a problem (which can be easy tested by themselves anyway). Then, they fix the issue after 1-2 days. Good luck fixing this yourself when your external load balancers have no public network connection, and new provisioned instances don't have it, too :)

- They changed instance types. The instances started months ago have a powerful CPU at 3.6Ghz which is shown in all their advertised benchmarks against DO/AWS/etc. Now they are offering 2.4Ghz CPU which is about 60% slower. When I told them about this I received an predefined template response like: "Thanks for your inquiry, we are constantly revising our hardware to meet customer needs".

- They increased pricing. First, the instance pricing was 20% lower than on DO. Then, they changed the price as beeing the same as DO and added a 20% discount. Then, they removed this discount, just telling everybody that they ended the discount period, nothing about the price increase :).

So, we are in the process of migrating everything to Google Cloud. We've been testing it on a staging cluster of 10 nodes and everything works smoothly, so we are in the process of moving the production cluster too. Be aware of small instance types (f1-micro and g1-small), they have very small CPU capacity for production load.

Also, Google offers us a lot of hosted stuff we don't need to manage anymore: DNS, Mysql, Datastore, Object store, etc. We were using Amazon S3 before, but peering from other datacenters (eg Vultr Datacenter, DO) sucks, constantly getting network timeouts, increased latency, etc.

The introductory could be done better, I agree. I think a full tutorial with code sample + lab account will help people to accelerate (I am slowly building up some code sample myself to train people internally). I suggest to AWS they need to slow down on making new products, and instead focus on enhancing the existing products. Some of the products lack of critical features, and in some cases their APIs are not complete. TODO TODO. Kind of embarrassing.

I can understand why they don't do that. Several reasons, 1) they assume people who are looking to build infrastructure in their cloud platform have some experience, if not, extensive experience with operation and building infrastructure. Second, there is a whole support model they sell. Lastly, they have partners to do training. IMO, AWS has done good job presenting intro materials to experienced operation folks in their reInvent videos.

I can go on to defend why I think the entry barrier exists, or how AWS can do better. Here is my only tip to any new AWS customers: your mileage is different so don' start plugging in a dozen AWS native services into your architecture just because everyone is using it. Start simple. In our case, we are forced to use two ELBs in front of our nginx and our router service, which eventually routes the requests to our app server. That's a lot of web servers passing traffic. There should be a native offering of Nginx and Apache IMO. ELB is garbage (I say this with a love and hate passion) if you want to replace Nginx / Apache entirely. Anyway, IMO, the major take away of building a cloud infrastructure is "instance is cheap to recycle [if you have the right tool]."

> This is where DigitalOcean and Vultr are profiting the most - publishing tutorials,

Are you referring to https://www.digitalocean.com/community/tutorials/ ? I find them useful, but some of the community tutorials are suggesting (maybe just me) wrong approach. I don't think DO is the first one, I remember browsing Linode content quite often back in the days.

Anyway, I think the gap is filled by developing a PaaS such as Herkou, on top of IaaS (AWS, GCE, Azure). Of course, you can build one yourself.

I see that aswell especially since most services aren't in every region yet. they mostly hang around the key regions.

The problem is that both Google Cloud and AWS have a much, much wider offer than Digital Ocean. This is the primary reason why DO can have a streamlined, simple interface while Google and Amazon need massive, opaque web dashboards.

I haven't used Google Cloud for long but it had a fairly nice UI actually. And they put their cloud SDK (command line tools) in the Ubuntu Partner repository which I appreciate, makes it almost as convenient as using OpenStack services.

Sure, but AWS don't exactly go out of their way to make themselves comprehensible. The need for an explainer just for their brandnames (https://www.expeditedssl.com/aws-in-plain-english) demonstrates the problem.

In reality, most brand names aren't very explanatory. Digital Ocean would be completely ambiguous if you weren't familiar with them. Azure? Could be anything. Hell, there was a time when people heard "Amazon" and wondered aloud why an online book store shared a name with the rainforest region.

The real problem is not the branding, it's the sheer amount of products. Frankly, the branding should be at the branch-level (Network/Servers/Databases/Data Analysis) rather than at the bottom leaf.

DO is not even close to be compared with AWS/GCE, the latter league, especially AWS, provides more sophisticated services that try to wrap some common design patterns behind, and present user an API for developer/programmer to interact with based their usage case.

DO is nice, and I use it for lightweight stuff like building a website, but only for that.

he meant DO beats them in design, documentation, UX. of course, AWS/GCE are more powerful.

This. I can use Datalab, and I can interact with everything, but there's clearly dozens of services and options lurking around the VM in the background that make me feel like I'm walking through a minefield.

Different markets, different business models. DO isn't going after large businesses. A startup could get traction on DO and then migrate to something like AWS when they need more flexibility but will have larger costs.

If DO started to offer things like load balancers, managed DBs as a service (similar to DynamoDB and RDS), and Storage (S3). That could change their business quickly. Somewhere between Heroku on AWS is the sweet spot IMO. Heroku is $$$ and very little to no flexibility and AWS almost has too many services now with no focus.

When it comes to the big money, cloud infrastructure providers aren't going to make much if they target the newbies. The folks with the money have the technical staff to begin with.

> they assume every potential customer have quite good technological knowledge

Their target market is large enterprises with massive IT departments staffed with people who "have quite good technological knowledge" - most startups/small companies have probably never fielded an unsolicited sales call from AWS while most enterprise IT staff are probably sick of hearing from them.

I think Azure is the one closest to have a simple service for web sites. I tried a free site with Node and it was really easy to get working. Azure also has a lot of the more advanced things that AWS and Google has which is very important for a lot of customers.

I think the AWS documentation is very good. It's not a simple platform though, there's a lot of complexity that comes with the flexibility and extent of their offerings.

Reply on news.ycombinator.com