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
I actually understand your point, I somehow feel the same way with Salesforce and a lot of other SaaS businesses. Why does salesforce charge me more per seat? I know their incremental cost of me adding a seat is close to zero...
Same can be said about most recurring/SaaS business models. What we are all (us selling recurring software licenses) doing is to spread the cost of running a development team across the usage of the code that the team develops. You can always discuss if CPU Cores is the right 'currency', and I am honestly not sure it is, however I think it is fair that those customers that use (and hence benefit most) from the code that my team produces should be those that pay most of the cost of that team.
Yes, we could go back to old-school perpetual licenses, but for most of the hosting/LET world that would not be realistic in terms of CAPEX. At OnApp we have +120 highly talented people at a cost well above $1m/mo, a perpetual license would have to be very very expensive to cover cost ongoingly.
At OnApp we feel that some sort of usage based billing is the right and fair model.
IDK for the Salesforce as I'm not familiar with it
And I actually can justify SAAS to some point - as they have expenses per customer / per site visitor (if someone run shopify shop and and some nice offer is shared on Groupon and its clones or in big facebook groups that can bring thousand of visitors to that site (even to couple of thousand online) for a day which will be a spike in usage and they got to serve that or fail as a service for $29 a month ...
but when solution is self hosted that expenses (per customer / per site visitor) do not exist for the software provider so price that charge that is nor relevant... (again there can be one unlimited price with which software providers is happy and cheaper limited prices - customers will decide if they like that price)
Intel was stagnating the IT industry with their CPU with 4-6 cores on a normal prices and super expensive multi cores 10-12 for about 8 years (v2 to v6 CPUs are not that different if there was not the hardware bugs which leads microcode patches which slowdown additionally the older ones... newer are just a bit more power efficient and not that much more powerful ...
Your prices was relevant for that period of time, but now days AMD start to glue cores... at first twice the number of cores than Intel with similar power and now days 4-6 times more and even faster than Intel ones so it is time to move forward ...
I do not know which is correct - but if 80% of the nodes of you customers are 4 or 6 cores you maybe can keep such fixed price at that price level... No one will complaint that price is not relevant to modern CPUs for which they pay the same price as for the 4-6 core Intel ones from couple of years ago and maybe introduce limited offers for older / smaller servers
You have to remember there's a cut off point - once you reach it, it's cheaper to hire in Openstack expertise and training than pay for licensing for things like Onapp, as there's no additional licence cost.
Problem with this per core model is that once you're racking 64 core compute nodes, the licence cost is ridiculous.
I think that in couple of years 64 cores will be under $200/mo market (my believe is based on that we now have $189 offer for 32 cores )
Which part do you disagree with?
That pricing should be usage based?
Or that pricing should be core based?
D
So, just some marketing bla bla and the same old shit pricing, nice.
Big issue is onapp is going to eat their own lunch with that.
Whatever they release is going to be a crippled full blown onapp.
Don’t expect clustered storage, high availability, etc. basically features people go to onapp for.
Francisco
I do not disagree I see it right if a license to have Onapp for brand new server with nice low-mid range $600-700 CPU was 40 bux in the past, (IDK your current prices as they are hidden) license for nice new low mid range server with $600-700 CPU should be $40 not 120 (Ryzen 3900) that is the way industry goes forward so maybe fixed price as this will be your future as in couple years with great probability nice midrange on that price will have 24 cores ....
I just download OpenNebula versus OpenStack - Competitive Pricing Review
for 64 cores nodes with premium level support will be on par for Canonical Open Statck even if the price is €2,5 ($3) per core even everything else is included for free
whoosh
Please see business model from their point of view to be sustainable and not just because you expect it. You're being irrational because of your minimal business skills.
Tl;dr you expect them to fix the price despite selling less and less licenses due to hardware being able to do more. Why you'd think their costs are fixed makes no sense to me.
Are we there yet?
In this business you pass the savings to customers - that is how it works or market makes you irrelevant 256 MB Vz VPS was $ 40/mo once upon a time - unmanaged or you add free stuff to the offer too keep it relevant (buy 10 nodes get 15 etc)
if a provider of the providers refuse to make that they are stagnating their customers and smart ones will start to drop them - first partially (switch storage to ceph at first) which will increase down spiral then one day when they bough or lease new hardware entirely (without notice) they will do new setup and migrate customers there ...
All big biz cloud software providers charge per CPU pair / Socket if the new secret price of Onapp is $6 per core the numbers are:
32 Cores x $6 per core ×10 nodes × 12 months is $23040 for a year add $100 for the controller and calculate storage as free
It is on par with Canonical Open Stack + Premium Support that includes OS support
and 50% more expensive than Open Nebula + Premium Support that includes OS support too
I do not think storage s is free with Onapp and I do not think that they include OS support
So Onapp is allready on cut off point for 32 cores with $6 per core for Open Stack + Premium support and if the price is $4.50 per core a bit over the price for Open Nebula + Premium Support
Thanks for your perspectives on all this. It's not a straight forward discussion, and for cloud service providers I do not think that OpenStack or Open Nebula are like for like deployments compared with OnApp. We are written specifically for hosters/CSPs and every feature we develop are build with you in mind. The same can not be said for OpenStack or Open Nebula, whom are more predominant with enterprise or large private cloud deployments.
Also, for this new initiative (re the original post) we are releasing we are introducing a fundamental switch in how pricing works, and how cloud platform is delivered. It's effectively a new category we are bringing to market. For one, the concept of 'cores' as a currency is gone.
All that being said, the points from both @coolice and @TimboJones are really good and relevant.
Oh, and:
OnApp storage is actually included in all our packages today.
D
What's the new currency then?
Why would that apply here? Those providers are spending the same amount and getting improvements others made and "passing on the savings" because they did fuck all and have no additional investment to recoup. Comparing to a service that has support and continually adding features makes no sense.
Well, if you're only considering the paper price and not the total cost of development and included features, you're not making a valid comparison. It doesn't matter if $service is cheaper if it's hot garbage. Generally, the expensive shit got expensive because it made/saved money for someone else besides the seller. Adding a human developer with skills is a 6 figure cost, so solutions with least amount of people could win comparisons that cost more on price alone.
OnAPPCoin.
Because no matter all bells and whistles that different software brings - for every client of the softwares - main business is to sell access to CPU, RAM, Space and main product the CPU for decades follows Moore's law... . Software just helps to sell it on a dedicated or fair share basis ... If you want to stay competitive the industry that followed the Мoore's law for so long you got to be able to constantly move forward...
If a Software provider in some way prevents theirs customers to follow their product lines with Moore's law they are stagnating their customers and make then noncompetitive....
If a software provider is unhappy how the things go with main product of their clients (Moore's law) they should go make software for some for some industry that do not have such product growth cycle (maybe accounting software or something else)
In terms of spend for development OpenStack and Open Nebula (has a lot more spend than Onapp ) cause of the main clients that use them and invest in that ... we should never talk about quality as some of the bigger and more respected providers run OpenStack
It is not big boys that positioned them to be the cheaper one - they are charging a lot per server node that do not make the calculation for providers with low core servers ... Onapp found their niche as the cheaper software to build a Cloud with cheap low cores server and this was ok for some time, but due to Moore's law an low core servers almost do not exist (we live in the time of cheap multi cores )... and this make them more expensive than big cloud software providers
P.S The only things it was missing with Big was easy auto provision modules but now days it is integrated in billings like hostbill has pretty decent Openstack Module
Are you a teenager or something? Partially about your inexperience, partially about the "..." , and partially how you don't understand hardware and software business.
Moore's Law is about the doubling of transistors, nothing to do with price of CPU, RAM and storage space.
These days, hardware is sold at low profit to sell software at high profit. It's been the gaming console business model for 40 years.
Hardware has margins like 10%, software like 65%.
Microsoft, Oracle, Apple, all software companies (Apple just found how to be profitable on hardware. Note the "designed in California", they don't MFG stuff, they have contractors do it). You might have heard the term "killer app" to describe an app that made a platform take off because it wasn't of interest before the killer app.
If their revenue stream decreases each year, what do you expect?
What? The company simply expects more money when more work is done. This is pretty fair. You're being irrational and/or cheap. You talk about Moore's Law, but what's the relevance when you're using ancient hardware and not following Moore's Law to 16, 32, 64 cores? You're not running server grade stuff, half of us had more powerful desktops 10 years ago.
They may have realized you guys and your small cores and 8 year old hardware isn't enough to run a business on. They'd rather sell to companies that do serious business and growing.
Moore's Law can be interpreted broadly as doubling CPU power ... no matter if it is single circuit or glued cpu cores (I like the glued ones - Intel at first make fun of AMD gluing x4 cores together with the infinity fabric but now days they do the same)
For others Industries ok about software they can profit as much as they can/ want.... in our industry software exist with one single purpose to help us sell access to the hardware to our customers....
If they care more about revenue stream than to help us to be as efficient as possible and keep up with market they just choose the wrong industry...
I do not believe a software company selling advance virtualization software is doing more work on their side when they sell exact same license to 32 cores EPYC then when License is used on 4 cores E3 ...
I mentioned before if a company doing "serious business and growing" is paying 180% of hardware price for software that helps them sell access to the hardware
And the completion choose to go open source (Opne Stack as big boys) or something cheaper they can literally offer 3 times more hardware resources with the same expenses at the same price for the end customer as you with same level of node density which will make you noncompetitive
Onapp can literally sense when a client of them is preparing to jump ship (maybe not immediately but on next hardware rotation) when they stop to use their storage even if it is now free if a client switch storage that is the first more important step as the new storage if it is supported can be attached other cloud setups and they just got to import VM configs to the new system and shutdown and reboot VMs there
P.S My cloud setup how I mentioned in the beginning is designed with addition of fail over with exactly the purpose to jump (5 min process) to new hardware without reinstalling shared hosting VMs which i plan to do 5 times in this decade so I'll keep moving forward out competing companies that pay licenses per core
Francisco
Jesus, no.
Buddy, YOU are in the wrong industry.
Hello, McFly? Learn what fucking licensing is for. Are you new? But you missed the point: YOU get more work done, means they provided more value.
/thread. Triggered by your abuse of the period.
They provide Exact Same Value for both servers (ability to sell access to hardware) + support + some bell and whistles, At the exact same cost for them
I think the correct licensing as unlimited for the node + some smaller ones if the provider want to attract such type of customers (Canonical and Open Nebula Agree with me as they licensed per physical server )
The pricing discussion > @coolice said:
I guess @coolice is proposing a bottom-up/cost-plus model, where license fees are based on incremental cost + a margin and @TimboJones is suggesting to look at a value based model where pricing is based on value created. Both makes sense, but none of them in pure-form.
We are going to market with a new model that will encompass both principles. I can't wait to show you.
Where neither of us will be satisfied. Lol j/k
LET'ers are not easy to satisfy
Not necessarily. The thing is - LET is one of the few places where you can have a conversation with your potential customers, without any brown nosing involved by either side.
absolutely!
LET is a tough crowd though, as they work on lowest margins in our industry though have high requirements to solidity and features of their software/platform partners.
If it works here it will work anywhere
Must admit we are certainly keeping an eye out for that email with news!!
What I'm wondering is, will it be enough to tempt us away from Openstack.
Do you want to be tempted away from it? Certainly, some do I guess.