Discussion in 'Business & Enterprise Computing' started by GooSE, Oct 4, 2013.
Here you go guys. IOPS scaling: http://youtu.be/B-RBDtKgQTo
This attitude from vendors shits me no end. At least with pricks like Cisco and Dell I can get a rough idea on pricing from public info. That means I can figure out if it's worth my while going through the pain of hitting up a sales goon for a better deal.
Reading between the lines, when someone says they can make it 'compelling' it also says to me that you'll try and screw me over if I'm not on the ball. Just give me the best deal you can straight up, don't play hard to get, I've been doing this asinine dance of the seven veils with vendors/resellers on pricing for too long and am heartily sick of it.
The cost of the HDS FC SAN and some more VM hosts with half terrabyte ram was so far under the base system costing from Nutanix, the Nutanix solution wasnt even in the ballpark.... which was a shame. It was new in April.
Agreed, hence my post in rants about Companies that don't provide pricing guides on their website.
Gives the price of a 4 node NX 3000 series as $144K, From that site, this seems to come with 5x1TB spinning disks and 700GB of SSD per node. The sizing guidlines say 400 VM's will run on this setup. (which assuming no disk redundancy, would work out at ~60GB/vm if spread evenly).
For something touted as being modular, it should be pretty straight forward to provide a price. I'm always skeptical of companies that are unwilling to discuss price publicly. I always read this as "Our base price + as much as we think you will be willing pay" which turns me off the vendor completely.
Disclaimer - I work for a vendor (specifically Nutanix) - The reason for this (not saying I like it or otherwise) is most vendors do business through the channel. Meaning, the Partner provides pricing to the customer not the Vendor.
If you want a quote, call up your trusted supplier, and ask them for comparative pricing on whatever solutions your considering. Prices vary for lots of reasons across the globe, so what a customer in USA pays may be different to an Australian customer for example, due to shipping, taxes, duty , exchange rates etc.
In the past, I always work out what my customer requires first (ie: technology/solution), then do pricing to see if its viable and then go back to the customer if the solution is outside their budget to discuss what requirements can change to get within the budget, or can they justify more budget to meet the requirements. Price (within reason) shouldn't be dictating the initial architecture.
First up I do not work for Nutanix. I would not discuss price in an open forum for any of the vendors I represent. Your response is part of the reason. You've clearly misunderstood my motivation for not doing so. In the case of Nutanix, there are may different models and configs. I can address your requirement in a number of ways. If we are trying to hit a price point, then we would not doubt use different nodes than if we were looking for ultimate performance. I need to understand your requirements. It's not a sales tactic, it's part of representing a solution no matter who the vendor is.
Let me know if there is a true interest, and I will send one of my 'Goons' to come see you.
Thanks, understood. I dont have much HDS experience and have only ever worked on a couple of projects with them so my knowledge of options, configs etc is limited. Nutanix as you appreciate is converged, so is compute and storage. Therefore, I see most traction when the requirement calls for both. A lot of customers talk to us when they want to refresh just storage, and in those cases Nutanix is not necessarily the best fit. It all depends on the requirements as I said in my other post. If there is a desire in the long term to get into a truly scalable, pay as you grow platform, then customers may transition on the back of just a storage requirement. If there is not, then they may not. I spoke to a customer at vForum in just this situation, and clearly from his requirements it's not a good fit. No one technology is the right fit for all requirements.
I work with a number of vendors, software and hardware and the number of sales at list price would be in single digits. I have seen traditional storage vendors slash pricing to suit their sales cycles and targets. Clearly you've experienced some of this in the past. I've always found that valuing and representing the customer, developing a relationship that allows us to bypass this song and dance brings the best long term relationships, and we have a number of them.
The NX-3000 can have either 800GB or 1.6TB of SSD per node. Again, part of the reason why forums are good for some things, but not others......
The solution is modular. Scales linearly for performance, capacity, and the all important price.
VMWare have historically over-promised and under-delivered on their product performance. Speaking for myself, I haven't ever been impressed with their price to performance ratio, which has always been the worst in any market they have a presence in.
VMWare's toolset is great, particularly if your IT staff aren't very bright. Everything is just a couple of clicks and you're done, which is good news for huge companies filled to the brim with mediocre staff. That's always been its selling point, and most definitely not the amazing performance they claim at vForum every year (dreams of which are quickly shattered as soon as anyone rolls this stuff out in the real world). Likewise you can always find a million "plugin" style products to do what you need (at dollar cost) rather than needing to write anything for yourself. Again, fantastic if your IT staff aren't very bright, and you don't want to do anything out of the ordinary or industry-specific.
For me, VSAN falls into the same boat. Looks great on paper, will be a piece of cake to install and configure, but I really have zero faith in any performance metrics VMWare are claiming, based entirely on being let down by their performance on everything else to date.
I understand that most sales do not complete at list price. But List price still exists, and is available as a starting point. If I'm researching potential solutions, I want to know at least ballpark figures, If you don't provide them I assume you are out of my price range (even though you may not be). and move on.
If you had the pricing information available, I wouldn't have googled 'Nutanix pricing' and received outdated results.
Before I worked at Vendors, It took me <1 day to get the pricing I wanted for these kind of comparisons, all I did was call my trusted vendor or partner.
I saw (and still see) no value in generic numbers on the internet as they are rarely anything like what I paid in the end and have no credibility. Better off just doing it right the first time. Even If I give you an exact price right now for product "X", it probably wont be correct in 7/14/30 days but that price will live forever on the internet creating confusion.
and RRP is always nothing like street price...
Understood. There are Nutanix guys on this thread, so I am sure they will see your feedback.
I'm pretty vendor skeptical, and am very much anti-push sales. I view "contact us for pricing" as a poor grab for my details so that you can have a high pressure sales guy ring me every week, and push me for details of which solution we ended up with, or if I had the numbers of anyone else who would be interested in their product.
Wow, I must not be too bright, I spent all this time working on business specific functionality running *on top* of the hypervisor, when I should have been designing the virtualisation infrastructure from the ground up to provide the absolute best possible performance/$.
I assumed that because a relatively minor investment in hardware easily covers our performance requirements that my time would be better spent doing things that actually benefit the users, what was I thinking?
Try not to insult people just because they happen to have different requirements to you, I've seen your other posts and know you are too experienced to actually believe that just because something works for you that it is therefore the one true answer to every problem.
My post covered two usage scenarios:
1) Staff who aren't very bright
2) Businesses that don't want to do anything very unusual.
If you consider yourself bright, I'm not going to argue. If VMWare does what you need it to do from a performance standpoint, then you fall into category #2 - ordinary business needs.
VMWare flat out does not meet the performance needs of the business I work for (media industry). It fails to deliver on IOPS and overall throughput across storage, networking and CPU on every test we've thrown at it. Competing technologies like Xen and KVM constantly put it to shame by a factor of 300% or more.
I used to work for a big boring finance corp who used VMWare. It worked wonderfully well for them - they could hire not very bright admins for low dollars and let them manage the VM infrastructure in a entirely standard way, and then they could hire much brighter people who spent their time up in the user-land applications doing very smart things. That worked for them.
Where I work now, every ounce of performance matters. The "relatively minor investment in hardware" statement doesn't work in media. When you require tens/hundreds of thousands of CPU hours to deliver media, every single ounce of performance matters. If software X is 10% slower than software Y, that can mean quite literally days of difference between delivery dates (or if the date is static, then the number of iterations of tests that can be done before the deadline, which has a direct result on quality).
So for you, a "relatively minor investment in hardware" means a moderately specced machine or ten running your hypervisor. For us, we're talking hundreds of render nodes that we buy at maximum spec (most are dual Xeons clocked at their max clockspeed, with a minimum of 128GB RAM each). We occasionally need to leverage virtualisation for some very tricky use cases, and loading VMWare with our media apps sees a massive performance hit across all types of operations. VMWare's standard response to their performance problems is "buy faster kit" - for us that's just ridiculous, when we're talking about nodes that are already highly specced, and performance we've lost courtesy of a virtualisation suite that cost good money to run with all the bells and whistles. Upgrading our infrastructure would cost an enormous amount of money just to prop up the performance of an already expensive piece of software.
Instead, we turn to open source hypervisors that give us far greater performance (often just a few percent shy of bare metal), and cost us nothing. We employ smart people to manage it, so trivial "point and click" type interfaces are moot - everything we do is done by folks who are borderline developers themselves, and we have substantial investment into tools like Puppet to do all the heavy lifting for us in a consistent manner with a fraction of the staff to admin (or machine to admin) ratios that you would see in a regular business.
So, are you wrong for buying a low-performing, commodity hypervisor that's easy to use? No. That works for your business, where clearly your investment needs to go into user-facing applications that don't really matter if they take a performance hit due to a relatively poor hypervisor (again, I've worked in finance and web development places where that sort of hit was never noticed by end users, because multi-core, multi-gigahertz servers are already much faster than required). For us, that doesn't work. it represents real-world cost to our business that's already facing a massive amount of competition in a cut-throat industry.
VMWare's "any idiot can use it" selling point is perfectly fine. That's not a criticism (and doesn't mean you're an idiot for using it). Their performance sucks. That is a criticism. If performance doesn't matter to you, then by all means go ahead and use the product. It matters substantially to us, so we'll avoid it like the plague.
For what it's worth, I sift through the Windows forums on OCAU, and constantly hear a barrage of Mac haters going on about how MacOSX's interface is for stupid people, blah blah. The irony being that there are some extremely brilliant people I work with on a daily bases who love MacOSX, and would run circles around most OCAUers. Likewise, I've met some very brilliant people who use VMWare. The choice of tool does not indicate the quality of the end user, regardless of whether or not the tool is designed to cater to the lowest common denominator. The two are not linked.
That's why you contact your local trusted solution integrator who is a partner of Nutanix who can give you the pricing you want.
We don't sell direct, so you don't have that issue with Nutanix.
Long story short, this way of doing business is nothing new, and the way most vendors do it rightly or wrongly.
No argument from me. But it still sucks.
If there was a better way, im sure it would have been discovered by now and vendors would do it.
Its ok to say something sucks, but you also should propose a solution which addresses the problems you think exist in the current method, without introducing more new issues.
IMO a big part of the reason why sales people ask you for your details is to
1. Avoid tyre kicking (a.k.a Time wasters)
2. Avoid competitors getting pricing and winning deals by discounting just below the price of the competitor
3. Avoid public (eg: Forum) discussions on pricing which may be misleading/outdated/incorrect which may impact reputation/business.
As a customer, if you want to avoid this, go through your trusted Solution Integrator who masks you from vendor sales people (if you dont like them) and work with them to decide what solution fits the requirements/budget, that's after all why they are there.
Look, all that's interesting for you, but as a customer I just simply don't care. My favourite vendors are the ones who post their pricing publicly, and don't force me to talk to crummy sales guys in order to figure out how much a rough rollout cost is going to be for some product.
Don't get me wrong, I empathise with your position. If I was in sales, I'd agree with all of the above. But I'm not. I'm just a crummy customer, and from that point of view I just don't care about the heartaches my vendors face when I'm too busy caring about the heartaches my department and greater business faces.
Potentially harsh, but 100% honest.
Oh, and no, I really don't need to address your problems. I've got my own to address. But let it be said that I won't be ringing your sales guy for a quote.
I've worked in channel in the past, and I doubt the sales tactics (and they are tactics) have changed enormously in the last few years.
Caterhamking, I'm having a really hard time interpreting your remarks as anything other than "don't you worry your pretty little head about the numbers, we know what's best for you", which is a rubs me the wrong way. It's never good when peoples start believing their own advertising.
The worst bit about having to talk to sales is that it blows out the amount of time it takes me to estimate project costs by at several orders of magnitude, simply because of the back-and-forth with each particular enquiry. Something I can figure out in 20 minutes with some simple modelling becomes days of phone calls and emails to people who don't understand or whose model/strategy/tactic is to hoard information, drip feeding it only where necessary. Multiply this by >1 supplier and even simple scoping exercises become excruciating.
Have the info on your website solves 1 & 3 - eg if I want to look at moving something to AWS/google/heroku/whatever I can do preliminary analysis very quickly. I wonder how much of their success is due to the ease with which people can get information from them.
As for 2, retail traders seem to cope with this phenomenon pretty well, as do the 'cloud' companies. If only telcos could be so accomodating. At least they have an excuse, as install costs can vary wildly depending on location.
Absolutely agreed 100%.
An average project for us would be a dozen vendors. Some more complex projects three times that easy. We waste so many man hours of people just sitting on phones talking to vendors, wasting their day.
And of course, they get a reseller or account manager, never a direct technical person. So then there's the back and forth between the AM and their technical people, as well as the back and forth with our guys. What should be 2 days of planning turns into 2 weeks in the blink of an eye. Our projects are crazy complicated, and the vendors are typically totally out of their depth when it comes to what we do (which is fine, because they're not in our industry, but their lack of supplied information up front means we need to go digging in the most painful of ways to get the info we need).
No surprises that when it comes to commodity items, the vendors that simplify their pricing models and provide pricing on their websites tend to get more work from us, even if they're a touch more expensive than someone else.
1/ Put the price, or if the price varies on implementation, some case studies (with prices) on your website. Then I wont need to waste your time (and mine) calling up for a quote
2/ Oh noes. Competition, what on earth are you going to do... If you don't have a price on your website, and your competition does, I read it as "we aren't confident enough in our product/service" and you don't even make it onto the shortlist of vendors.
3/ Put the price on your website, that way, if someone says "It cost my A Gorillion dollars and was shit" someone else is likely to say "Hey look at the official website, its nowhere near that much"