vSphere 6 released today!

Mabrito

Supreme [H]ardness
Joined
Dec 24, 2004
Messages
7,004
A surprise in my inbox this morning...vSphere 6 is official today. I knew it was in beta and that VMware was doing a 28 days of February thing...but I still figured this would be official in August when VM World happens.

I for one am excited for it...Fault Tolerance finally gets past the 1 vCPU limitation...centralized library, VSANs, and improvements to NSX. Should be a good release!
 
I thought it was a bit of a snoozer myself.

I don't use FT anywhere, and more vCPUs won't change that.

Despite them repeatedly beating the " hypervisor is not a commodity" drum, this release just illustrated to me that we're getting there, except perhaps for the largest environments.

NSX .. 400 customers. vSAN .. 1000 customers. Virtualize SAP HANA.. How many customers there?

That said, still excited to get my hands on it and have a web client that's supposedly not molasses slow. Maybe I just need to get a job in a larger environment to touch some of this other stuff. ;)
 
I'm actually quite impressed at how quickly SDN and NSX especially are being embraced by the industry when you consider how entrenched legacy network architecture is. NSX adoption is faster than ESX was by a large margin and many of those 400 customers were multi-million dollar deals.
 
Last edited:
For me it's hard to get excited about v6. Doesn't bring any features/enhancements that will have an impact on our bottom line, doubt we will upgrade to it this calendar year.

NSX adoption is faster than ESX was by a large margin and many of those 400 customers were multi-million dollar deals.

Ok, let's re-read what you wrote. NSX adoption is faster than ESX? I'd ask you to cite your source, but your claim is so wildly unrealistic that no source is required.

NSX solves a niche problem. Yes, there are some customers who will realize a tremendous benefit from NSX, but the overwhelming majority of those who run hypervisors will have exactly zero use for NSX.

According to the marketing materials "provisioning network services is complicated stuff" that NSX will make oh-so-much better. Yes, it's complicated if you are clueless, or if you need to do it 1,000 times per second, but either of those is a problem few businesses actually have.

Again, according to marketing materials "eBay, AT&T, and Rackspace reduced provisioning times", whoop-dee-doo. America, heck, the World, runs on small to medium businesses, not on AT&T sized businesses.

I am not dissing NSX for what it is, all I am saying is that SDN is a solution to some people's problems, but it will take a long time for it to have the same market penetration as a hypervisor which at this point in time is ubiquitous.

Back to v6; what I see is that the hypervisor is indeed becoming a commodity, and that's good for consumers as hypervisor providers will be forced to compete solely on price and quality of support.
 
My source is a briefing from our internal 2015 kickoff last week. I will ask if I can share the details and I will try to address the rest of your NSX points tomorrow.
 
Regarding your niche problem statement, do you consider security a niche problem? And yes provisioning times are an issue for almost every customer I talk to and they are all taking a hard look at NSX. Yes it will take time but the writing is on the wall. Compute resources are spun up in minutes then wait days sometime weeks for the 90's style network operational model to catch up. And it takes even longer (if it happens at all) for said network resources to be rolled back once the compute resources are no longer needed.
 
Last edited:
Regarding your niche problem statement, do you consider security a niche problem? And yes provisioning times are an issue for almost every customer I talk to and they are all taking a hard look at NSX. Yes it will take time but the writing is on the wall. Compute resources are spun up in minutes then wait days sometime weeks for the 90's style network operational model to catch up. And it takes even longer (if it happens at all) for said network resources to be rolled back once the compute resources are no longer needed.

While I would love to get my hands on NSX, our old style networking doesn't take long at all. We use rancid and have scripts for just about everything. Provisioning is just a matter of entering in the names and vlans. De-provisioning is also done the same way.
 
How do you expose that to users? How many man hours does it take to build and maintain?
 
Last edited:
How do you expose that to users? How many man hours does it take to build and maintain?

Hey, I didn't say it was all automated. ;) This is internal use only at a small company. Only users using it would be the ones handling provisioning anyway. I am working on vCO to bring even more of it together. Since vCO integrates with AD, ssh, powershell, etc. It can be pretty powerful and is technically free. Sure it takes time to build the scripts just like most scripting does. But ones it is complete, it is easy to use and it you keep your scripts clean and organized, it is not difficult to add additional switches, ports, etc in the the future.

I want to get into NSX and bug management all the time about it, but it is not exactly cheap.
 
How do you expose that to users? How many man hours does it take to build and maintain?

We have a 4 individual contributor network team that maintains telephony (~85% VoIP, growing) and LAN/WAN networking for a company of ~9k employees with 2 data centers, 7 major geographically distributed regional offices and about 10 local offices off of the 7 regional ones. We are a 24/7 operation and lives could literally be lost if IT screwed up.

It doesn't take our people weeks or days to provision any network resources, in addition our network topology doesn't constantly change.

There are absolutely use cases for SDN where SDN provides a tremendous value. It's just like VDI, there's a use case for it but it's not nearly as ubiquitous as hypervisors because the SDN or VDI cost/benefit just isn't there for large segments of the market.
 
I just hope I have time to learn vSphere 6. I really wanted to get my VCAP5 probably during spring, but I also didn't think vSphere 6 was going to be released in Q1.
 
We have a 4 individual contributor network team that maintains telephony (~85% VoIP, growing) and LAN/WAN networking for a company of ~9k employees with 2 data centers, 7 major geographically distributed regional offices and about 10 local offices off of the 7 regional ones. We are a 24/7 operation and lives could literally be lost if IT screwed up.

It doesn't take our people weeks or days to provision any network resources, in addition our network topology doesn't constantly change.

There are absolutely use cases for SDN where SDN provides a tremendous value. It's just like VDI, there's a use case for it but it's not nearly as ubiquitous as hypervisors because the SDN or VDI cost/benefit just isn't there for large segments of the market.
I think of SDN as moving networking services into the hypervervisor and giving users a unified control plane with full context and APIs to consume those services. Obviously some will benefit more than others and I was surprised at how responsive the market has been on the distributed firewall alone.
 
I think of SDN as moving networking services into the hypervervisor and giving users a unified control plane with full context and APIs to consume those services. Obviously some will benefit more than others and I was surprised at how responsive the market has been on the distributed firewall alone.

The distributed firewall is one of my biggest reasons for wanting NSX. I hate to admit it, but we use pfSense for all our virtual firewalls. At least with FT I can have it running on more than one server. And luckily vSphere 6 now supports more than 1 vCPU as we have been testing the latest pfSense version with 2 vCPU. It is hard to argue the cost benefits of moving away from a free firewall.
 
The distributed firewall is one of my biggest reasons for wanting NSX. I hate to admit it, but we use pfSense for all our virtual firewalls. At least with FT I can have it running on more than one server. And luckily vSphere 6 now supports more than 1 vCPU as we have been testing the latest pfSense version with 2 vCPU. It is hard to argue the cost benefits of moving away from a free firewall.
pfSense supports CARP, you don't need to use FT.
 
pfSense is great as a simple cheap edge firewall, I use it at home and I really like it. NSX DFW is not an edge firewall, NSX Edge GW is used for NAT, VPN, etc...
 
For me it's hard to get excited about v6. Doesn't bring any features/enhancements that will have an impact on our bottom line, doubt we will upgrade to it this calendar year.



Ok, let's re-read what you wrote. NSX adoption is faster than ESX? I'd ask you to cite your source, but your claim is so wildly unrealistic that no source is required.

So when Pat Gelsinger said himself plus the NSBU, who I had dinner with last night, said it again they were lying? People were very skeptical of ESX back in the day..just like NSX now. But NSX is taking off VERY quickly.
 
So when Pat Gelsinger said himself plus the NSBU, who I had dinner with last night, said it again they were lying? People were very skeptical of ESX back in the day..just like NSX now. But NSX is taking off VERY quickly.

This just has to be one of those Darrell Huff How to Lie With Statistics moments. I absolutely believe the "2x sales growth compared to ESX back in the day" claim if he is referencing ESX Server version 1.5 from like 2000 or whatever it was which they probably sold like 1 a month or some such.

Let's just put all the marketing and "visionary quadrant" crap aside and look at the numbers.

How many entities have adopted NSX and actually paid for it?
How many entities run ESXi?

Compare the answers for those two questions to each other and there's your answer for whether NSX (and by extension SDN) adoption is sweeping the industry or not.

IMHO this isn't like ESX back in the day when virtualization was unproven technology. I think it will be hard to find anyone who would deny that SDN is the future.

It's very much a question of whether the NSX cost is worth the benefit it provides. For some entities the answer is YES. For most (of those who run ESXi) it's NO at this time.
 
Regardless of its adoption rate, despite how excited I am about NSX the product and technology I'm approaching NSX cautiously. No doubt it's a fantastic product and will enjoy a lot of success, but I like to see VMware products, updates, patches, etc. bake in the wild a while before I start to encourage people to put them in production.

Now that I have NSX in the lab and see more how it works first hand, I'm gaining a lot more comfort and confidence with it. It's cool technology.
 
Regardless of its adoption rate, despite how excited I am about NSX the product and technology I'm approaching NSX cautiously. No doubt it's a fantastic product and will enjoy a lot of success, but I like to see VMware products, updates, patches, etc. bake in the wild a while before I start to encourage people to put them in production.

Now that I have NSX in the lab and see more how it works first hand, I'm gaining a lot more comfort and confidence with it. It's cool technology.

what lab do you have it in? a home lab? VMware quoted us 100k$ last month to put it in our lab, and told us it REQUIRES a PSO engagement, no doing it ourselves. NSX is cool stuff, but short of fortune 1000 companies, i dont see anyone implementing it for the next 3+ years. and they appear to be intentionally pricing it out of range of anyone else.
 
what lab do you have it in? a home lab? VMware quoted us 100k$ last month to put it in our lab, and told us it REQUIRES a PSO engagement, no doing it ourselves. NSX is cool stuff, but short of fortune 1000 companies, i dont see anyone implementing it for the next 3+ years. and they appear to be intentionally pricing it out of range of anyone else.
Some partners have access to the bits. As for PSO required, I would need to see the details of the proposal but it sounds like you're talking about a paid POC, the cost of which would vary on based on the level of effort. You can definitely buy NSX and DIY, I've had a couple customers start with a small 4-8 socket purchase with the intention of expanding as needed.
 
Any benefits for a ESXI Home server environment?
How have upgrades gone for smaller environments gone?
 
Some of us have only had the beta, not GA release yet. Nothing really new for home environments in way of basic virtualization/AIO setups.
 
Regardless of its adoption rate, despite how excited I am about NSX the product and technology I'm approaching NSX cautiously. No doubt it's a fantastic product and will enjoy a lot of success, but I like to see VMware products, updates, patches, etc. bake in the wild a while before I start to encourage people to put them in production.

Now that I have NSX in the lab and see more how it works first hand, I'm gaining a lot more comfort and confidence with it. It's cool technology.

It's epically awesome tech, that's for certain, but it is priced at the "slightly nutso" range for most places right now. To be honest, that's how they want it (I won't go into the reasons; I disagree with them, but it is what it is).
 
It's epically awesome tech, that's for certain, but it is priced at the "slightly nutso" range for most places right now. To be honest, that's how they want it (I won't go into the reasons; I disagree with them, but it is what it is).

If VMware was serious about these products.. they would offer the average Joe user trial lics for its NSX and VSAN products just like they did with ESX

Like millions of people out there I became invested in ESX because I had my free 90 day trial lic. Then I took that experience with me and showed the crappy company who ended up coughing up the moneys
 
If VMware was serious about these products.. they would offer the average Joe user trial lics for its NSX and VSAN products just like they did with ESX

Like millions of people out there I became invested in ESX because I had my free 90 day trial lic. Then I took that experience with me and showed the crappy company who ended up coughing up the moneys

Except that's ~not~ what they want this time. This is the part I disagree with, so we'll leave it at that, but I ~entirely~ agree with you.
 
If VMware was serious about these products.. they would offer the average Joe user trial lics for its NSX and VSAN products just like they did with ESX

It's about the anchor point for pricing.
Right now there are some enterprises out there who aren't flinching at paying millions to license NSX across the enterprise because it actually saves the enterprise money.

In order to maintain that price point, and the tremendous returns the NSX licensing generates it's imperative to make the early adopters pay through the nose.

Once the sales in the Fortune 100 slow down the pricing will be adjusted to make it attractive for Fortune 500, then 1000, then 5000, and at some point Joe will be able to purchase NSX for a price he can afford for his small business.

There's no point in making NSX accessible for the masses at this time because the margins are just too good selling to the global enterprise customers.
 
There's no point in making NSX accessible for the masses at this time because the margins are just too good selling to the global enterprise customers.

As an NSX customer, it is of my opinion that the hesitation to offer this to the masses is a lack of support ability at a broader of level. If it's not a completely green field environment, NSX in its current state offers a lot of unexpected complexity when things go wrong.

Often times, the support engineers we work with seem to be largely over-burdened with support calls. Not to detract from the level of service, as they've all been extremely knowledgeable and very responsive - just very apparent they are in low supply and high demand.

The product does not seem to be ready for the average vmware customer to purchase and deploy with minimal frustration. As such, probably a good idea it's priced out of reach until that bit is sorted out.
 
The product does not seem to be ready for the average vmware customer to purchase and deploy with minimal frustration. As such, probably a good idea it's priced out of reach until that bit is sorted out.
I think it's not necessarily the product that is not ready and this is just my opinion. The traditional VMware customer is the server infrastructure team and often don't manage the network or have the expertise to do so. As I speak with customers about NSX they want to know what part of their staff will manage it. Traditional network engineers often have a hard time embracing virtualization and handing some responsibility to vSphere admins they don't trust. There are some organizational challenges to be worked out before NSX can just be downloaded and plugged in like ESX was.

That said, the changes are happening and almost everyone I talk to is interested, the DFW alone is compelling enough for people to take a hard look at how they run their network.
 
So, to me it looks like there is next to no difference for us humble home users using the free edition.

We are still stuck with the client, and the client is still stuck at the 5.0 feature level (though it supports version 11 guests, just greys out features newer than v8)

I was really hoping 6.0 would give free users more teaming options like in the licensed 5.5, instead of being stuck with a single static "route based on IP hash" group per vswitch.

Some good news at least. 3rd party drivers compatible with 5.5 will still be compatible with 6 (though might be tougher to install) (Wooohoo, get to keep my brocades).

As it stands right now, I'm trying to determine if I have any reason to upgrade at all once it hits. Maybe there are some security/stability patches that make it worth it?

Also, for a free version user, are there any advantages to running v11 guests over v8 guests if all the new features are disabled anyway?
 
Last edited:
Back
Top