AWS Migration tools? Anyone with VMware on prem and moving to AWS?

nicholasfarmer

Limp Gawd
Joined
Apr 19, 2007
Messages
238
Assuming the old HardForum Crew still watches these forums...

Has anyone, with VMware in a datacenter, started, has started, already completed... a project to migrate workloads into AWS?
Google search does turn up some blogs and a list of tools.
AWS - at enterprise scale - does come with consultants and help, but when you hit scale and actively make the move, hands on experience is priceless.

Any thoughts or comments are very welcome, please!
 

socK

2[H]4U
Joined
Jan 25, 2004
Messages
4,016
If you are just planning on doing a lift and shift, you're missing some of the best perks of the cloud - and if cost is an issue, you will violently exasperate the issue by attempting it. Just dumping VMs in AWS is going to be expensive, and doubly so if you really feel like moving your VMware environment up there. It's the most work to re-engineer an application to run up there, but it'll be the most robust and cheapest by far.

Say you want to migrate a web server. Bring it up in an ECS container, keep static content in S3, use RDS for your database.
Container dies? Let an auto-scaling group respawn it, it's basically stateless and all data lives elsewhere. Need actual resilience? Keep multiple containers alive and attach them to a load balancer. Want to patch it? Kill the container and it let it come back up.

Have simple utilities or scripts that run off cron jobs? Move them over to Lambda.

Maybe you can transition to some of their managed services and just outright get rid of some instances you run. Depends on what you value your time at. RDS has an obvious price premium over running the database yourself... but things like backups, restores, patching, storage, multi-region, HA, largely becomes trivial or someone else's issue.
 

nicholasfarmer

Limp Gawd
Joined
Apr 19, 2007
Messages
238
If you are just planning on doing a lift and shift, you're missing some of the best perks of the cloud - and if cost is an issue, you will violently exasperate the issue by attempting it. Just dumping VMs in AWS is going to be expensive, and doubly so if you really feel like moving your VMware environment up there. It's the most work to re-engineer an application to run up there, but it'll be the most robust and cheapest by far.

Say you want to migrate a web server. Bring it up in an ECS container, keep static content in S3, use RDS for your database.
Container dies? Let an auto-scaling group respawn it, it's basically stateless and all data lives elsewhere. Need actual resilience? Keep multiple containers alive and attach them to a load balancer. Want to patch it? Kill the container and it let it come back up.

Have simple utilities or scripts that run off cron jobs? Move them over to Lambda.

Maybe you can transition to some of their managed services and just outright get rid of some instances you run. Depends on what you value your time at. RDS has an obvious price premium over running the database yourself... but things like backups, restores, patching, storage, multi-region, HA, largely becomes trivial or someone else's issue.
Preaching to the choir... trust me.
I'm fully versed in what we 'should' do and I wish I had the weight to push things in the right direction.
We already have a mature Kubernetes environment and entire teams 'cloud ready' with Azure, Google, AWS workloads.
Best I can do atm is learn the 'migration' options so I can give correct information when asked.

The question this week is attempting to deal with a ton of brown field workloads. Some that have app teams that no longer understand the application enough to convert it or support it outside of its current bloated virtual machine. I'm sure we can deal with the basic tools, Scan the environment, find good candidates, migrate/replicate, burn money?
Migration Evaluator, Migration Hub, CloudEndure?
VMware on AWS?
Any third party tools that reduce the drama that are worth the admin overhead to stand up and maintain the tool?

Ultimately looking for 'stay away from this' but most wouldn't know it until you are deep in that bit bucket.
 

socK

2[H]4U
Joined
Jan 25, 2004
Messages
4,016
There is VMware Cloud on AWS (amazon.com) but given that it's a "contact us," expect it to be expensive. Internally it's probably multiple bare metal instances dedicated to you and I'd guess when it's all said and done, nothing would really be stopping you from just adding some resulting cluster to your vSphere and moving an instance over. I'd wager you're looking at, at least $10k a month. I haven't used it, couldn't tell you for sure.

There will probably be some networking hoops to jump through, I'd assume you'd need to bring the thing up in a VPC with a site-to-site VPN or Direct Connect circuit, then some additional bits to get DNS working across the networks.
 

nicholasfarmer

Limp Gawd
Joined
Apr 19, 2007
Messages
238
There is VMware Cloud on AWS (amazon.com) but given that it's a "contact us," expect it to be expensive. Internally it's probably multiple bare metal instances dedicated to you and I'd guess when it's all said and done, nothing would really be stopping you from just adding some resulting cluster to your vSphere and moving an instance over. I'd wager you're looking at, at least $10k a month. I haven't used it, couldn't tell you for sure.

There will probably be some networking hoops to jump through, I'd assume you'd need to bring the thing up in a VPC with a site-to-site VPN or Direct Connect circuit, then some additional bits to get DNS working across the networks.
VMware Cloud on AWS is expensive. Almost 5x the cost of 'doing it right' in AWS.
You can install appliance VMs to pull L3 networks between the datacenter and AWS but you have latency and trombone issues...

VMware VMC provides that 'no change' migration to AWS at a cost. The simple vCenter GUI is awesome for companies that do not want to learn/train a new AWS based GUI, blue prints, security policies, application groups etc...
I have a feeling the cost is going to push against the VMware option.

In terms of POC.. Vmware will not give access unless you have a proven application 'cloud ready' for the POC. The document you fill out for the POC is large.

Thanks for the feedback, its helpful.
 

pizzahut4life

Weaksauce
Joined
May 10, 2008
Messages
87
I would recommend Terraform. It will help you manage your infrastructure in the cloud as code. I've only used it for a week or two, but the capabilities are vast.

It has an import option so you can import your vmware environment state. Could be helpful.
 

Shockey

2[H]4U
Joined
Nov 24, 2008
Messages
2,054
Zerto is competitor to Dell recoverpoint and VMware site recovery manager. It’ll replicate VMs to most major cloud platforms.

licensing is based on number of VM protected.
 

somebrains

[H]ard|Gawd
Joined
Nov 10, 2013
Messages
1,411
VMware Cloud on AWS is expensive. Almost 5x the cost of 'doing it right' in AWS.
You can install appliance VMs to pull L3 networks between the datacenter and AWS but you have latency and trombone issues...

VMware VMC provides that 'no change' migration to AWS at a cost. The simple vCenter GUI is awesome for companies that do not want to learn/train a new AWS based GUI, blue prints, security policies, application groups etc...
I have a feeling the cost is going to push against the VMware option.

In terms of POC.. Vmware will not give access unless you have a proven application 'cloud ready' for the POC. The document you fill out for the POC is large.

Thanks for the feedback, its helpful.
I was at the AWS Loft in SF last year with the #5 & #6 vcdx’s for a weeklong Sec event.
Cost isn’t what you focus on in any migration or you’re clinging to long arc CapEx and TCO stumbling blocks instead of focusing on agility.

1:1 lift and shift is not what any of the vendors are interested in.
You’ve seen that with the POC docs.
That’s just helping them trumpet some customer success and they can hammer a few bullet pts to sell to the masses that don’t do much homework.

Think about what needs to be pushed 1:1 as the applications that take too long to rearchitect, hopefully that pool is small.
Now the other applications can be pie in the sky with the teams that own them.
That’s where you start with yesterday and start running what if’s 1-2-3-4-5 years down the road arcs with them and see what services you can adopt to breech the “I want to use xxx but don’t want to support yyy to achieve it”.

What do those teams want to deprecate in favor of say 5 stacks you end up dropping at the end of a month is a lot of freedom and agility to throw at teams that aren’t used to it.
 
Top