Thin clients?

[aqt4u2lv]

Sun, Dell, HP/Compaq and Microsoft with XP embedded for thin clients.

What's the deal?


They still cost like $500 like a cheapo dell desktop.



I always thought having all the users logging on and running things locally on the server was just asking for them to screw the server up?

 

[lansalot]

Not if you tie down their permissions. It's one location to have an application patched and up to date, users can't install software on it, etc.

There's a long list of good reasons - but as usual, it all depends on your wants and needs.

 

[spidey07]

actually there are many large enterprises that use thin clients.

Face it, they are just incredibly easy to support and you can justify your return.

Let's say a company has 10000 PCs at 1K each.

10K * 1K = 10 M

Hardware refesh of 2 years. 10 M every 2 years.

Desktop support budget of maybe 2M (100K per employee or estimated outsource) per year.

Switch to thin clients plus your support cost would go down to less than 1 M.

you get the picture?

 

[lansalot]

Also, I should have mentioned: re "having all the users logging on and running things locally on the server was just asking for them to screw the server up".

Give them, for example, a shortcut that directs them into the published application straight away; ie - no explorer, no windows desktop. Close the application? You are logged off the server immediately. No way round it.

 

[aqt4u2lv]

Originally posted by: lansalot
Also, I should have mentioned: re "having all the users logging on and running things locally on the server was just asking for them to screw the server up".

Give them, for example, a shortcut that directs them into the published application straight away; ie - no explorer, no windows desktop. Close the application? You are logged off the server immediately. No way round it.

lol I'm talking about windows

 

[lansalot]

So am I. That was his quote, not mine 🙂

 

[n0cmonkey]

I love SUN's thin clients. They're sexy. I would love to have them setup in my house.

The benefits of patching a couple of servers as opposed to a few hundred desktops seems worth while to me.

 

[spidey07]

Originally posted by: n0cmonkey
I love SUN's thin clients. They're sexy. I would love to have them setup in my house.

The benefits of patching a couple of servers as opposed to a few hundred desktops seems worth while to me.

now think larger scale. 10s of thousands of desktops.

 

[aqt4u2lv]

Originally posted by: spidey07

Originally posted by: n0cmonkey
I love SUN's thin clients. They're sexy. I would love to have them setup in my house.

The benefits of patching a couple of servers as opposed to a few hundred desktops seems worth while to me.

now think larger scale. 10s of thousands of desktops.

Server dies, 10,000 employees are disabled

now how much revenue will be lost waiting for repair?

 

[Rogue]

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: n0cmonkey
I love SUN's thin clients. They're sexy. I would love to have them setup in my house.

The benefits of patching a couple of servers as opposed to a few hundred desktops seems worth while to me.

now think larger scale. 10s of thousands of desktops.

Server dies, 10,000 employees are disabled

now how much revenue will be lost waiting for repair?

Nice thought, but not based on reality or the common sense of any good IT person. Naturally, you would have redundancy and failover built into anything that runs the core of your business. No jackass is dumb enough to place 10,000 users on a single point of failure and if he is, he'll be fired in short order the first time it goes down. Nice try though...NEXT!

 

[n0cmonkey]

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: n0cmonkey
I love SUN's thin clients. They're sexy. I would love to have them setup in my house.

The benefits of patching a couple of servers as opposed to a few hundred desktops seems worth while to me.

now think larger scale. 10s of thousands of desktops.

Server dies, 10,000 employees are disabled

now how much revenue will be lost waiting for repair?

I don't know about the others, but 1 SUN server doesn't typically handle 10k thin clients. 😛

 

[spidey07]

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: n0cmonkey
I love SUN's thin clients. They're sexy. I would love to have them setup in my house.

The benefits of patching a couple of servers as opposed to a few hundred desktops seems worth while to me.

now think larger scale. 10s of thousands of desktops.

Server dies, 10,000 employees are disabled

now how much revenue will be lost waiting for repair?

there are load balancers and clustering that make that just about impossible.

In high availability computing you can take down any server for whatever reason without disruption in service.

Typicaly with thin client applications you have a large cluster of servers serving. From 4-50. If one goes down it isn't even noticed other than a blip on the operations screen of "server cluster134 is unavailable because of xxxx, servers cluster1-cluster133 are operational"

It all comes down to managability. What's easier? 10K desktops or a few hundred servers housed in key data centers across the globe?

 

[spidey07]

Originally posted by: n0cmonkey

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: n0cmonkey
I love SUN's thin clients. They're sexy. I would love to have them setup in my house.

The benefits of patching a couple of servers as opposed to a few hundred desktops seems worth while to me.

now think larger scale. 10s of thousands of desktops.

Server dies, 10,000 employees are disabled

now how much revenue will be lost waiting for repair?

I don't know about the others, but 1 SUN server doesn't typically handle 10k thin clients. 😛

True.

But a single 10K running an app for 10K clients is in norm.

 

[WiseOldDude]

You can have less processing power at the desktop which means less $$$, and you don't have to have all the different applications on the desktop, just the thin client, again saving $$$

 

[aqt4u2lv]

What would you use to cluster?

I'm talking microsoft solutions

 

[spidey07]

Originally posted by: aqt4u2lv
What would you use to cluster?

I'm talking microsoft solutions

holy mackeral....

depends on what you're trying to do and what the applications are.

me personally, I would use layer7 switches into a large server farm.

 

[aqt4u2lv]

Originally posted by: spidey07

Originally posted by: aqt4u2lv
What would you use to cluster?

I'm talking microsoft solutions

holy mackeral....

depends on what you're trying to do and what the applications are.

me personally, I would use layer7 switches into a large server farm.

uhh I'm talking about the server


The software has to know what's going on on the different servers

 

[WiseOldDude]

Originally posted by: aqt4u2lv

now how much revenue will be lost waiting for repair?

Clustering, a fault tolerant strategy. If one server fails, everything it was running "failover" to other servers in the cluster.

And truthfully your concern of a failing server also has the same effect on client/server environments. The server is still a central repository for all the data, database, with customer records, orders, etc., etc., etc.,

The network that is most likely to be completely down due to ?server failure? is the small company such as a law office, or other business that is dependent on the data and services provided by the server.

 

[spidey07]

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: aqt4u2lv
What would you use to cluster?

I'm talking microsoft solutions

holy mackeral....

depends on what you're trying to do and what the applications are.

me personally, I would use layer7 switches into a large server farm.

uhh I'm talking about the server


The software has to know what's going on on the different servers

no it doesn't.

We're talking terminal servers here. With a layer7 switch and even the clustering services themselves it doesn't matter.

You can have literally dozens to hundreds of servers all serving the same thing and if the server farm/network is setup correctly losing a front end server is immaterial.

We're talking terminal servers here...the front end. The middle end has to maintain state of course and from there the back end is just a couple huge fricking DB servers.

Take a look at 3 tier architectures and hardware assisted load balancing (layer 7/app switches) and it will become clear.

 

[aqt4u2lv]

Originally posted by: spidey07

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: aqt4u2lv
What would you use to cluster?

I'm talking microsoft solutions

holy mackeral....

depends on what you're trying to do and what the applications are.

me personally, I would use layer7 switches into a large server farm.

uhh I'm talking about the server


The software has to know what's going on on the different servers

no it doesn't.

We're talking terminal servers here. With a layer7 switch and even the clustering services themselves it doesn't matter.

You can have literally dozens to hundreds of servers all serving the same thing and if the server farm/network is setup correctly losing a front end server is immaterial.

We're talking terminal servers here...the front end. The middle end has to maintain state of course and from there the back end is just a couple huge fricking DB servers.

Take a look at 3 tier architectures and hardware assisted load balancing (layer 7/app switches) and it will become clear.

Dude, I'm talking about the software.



If you are logged into one server running word and someone is on another running something else it won't know about just because you have some magic switch.

 

[spidey07]

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: aqt4u2lv

Originally posted by: spidey07

Originally posted by: aqt4u2lv
What would you use to cluster?

I'm talking microsoft solutions

holy mackeral....

depends on what you're trying to do and what the applications are.

me personally, I would use layer7 switches into a large server farm.

uhh I'm talking about the server


The software has to know what's going on on the different servers

no it doesn't.

We're talking terminal servers here. With a layer7 switch and even the clustering services themselves it doesn't matter.

You can have literally dozens to hundreds of servers all serving the same thing and if the server farm/network is setup correctly losing a front end server is immaterial.

We're talking terminal servers here...the front end. The middle end has to maintain state of course and from there the back end is just a couple huge fricking DB servers.

Take a look at 3 tier architectures and hardware assisted load balancing (layer 7/app switches) and it will become clear.

Dude, I'm talking about the software.



If you are logged into one server running word and someone is on another running something else it won't know about just because you have some magic switch.

um, yes it will. That's the whole purpose of load balancing.

I sincerely suggest you read up on high availability computing and what can be accomplished. Think outside of single client to single server and dig deeper to the actual state of a connection.

That's the whole purpose of a "server farm"

 

[irwincur]

Also remember, that $500 for a PC (thin client) is cheap for a business. Yes a normal PC may cost that much, but then add warranty support, upgrades, etc... It adds up quick, and companies get hit by the extras. Even a good deal for a large business is hard to keep under $1200 or so. The PC companies know who has the money, and they get it one way or another.

Not to mention, thin clients really let you cut down on costly IT requirements. Staff is expensive, double your salary and that is the base cost to the company. So even an average help desk person is costing anywhere from $60000 to $80000 depending on the market. Not cheap. Now add a full staff of networking experts and admins to that. With a huge server farm you can cut staff requirements by more than 1/3. Over the course of a few years, that alone saves enough for the added cost of redundancy.

Plus, there is another growing trend of small and mid sized companies that do not want and can not afford an IT staff. Solution, thin client connections to an offsite, backup up, and redundant leased server. It will provide them with 10x more reliability than an in house IT staff.

 

[spidey07]

Originally posted by: irwincur
Also remember, that $500 for a PC (thin client) is cheap for a business. Yes a normal PC may cost that much, but then add warranty support, upgrades, etc... It adds up quick, and companies get hit by the extras. Even a good deal for a large business is hard to keep under $1200 or so. The PC companies know who has the money, and they get it one way or another.

Not to mention, thin clients really let you cut down on costly IT requirements. Staff is expensive, double your salary and that is the base cost to the company. So even an average help desk person is costing anywhere from $60000 to $80000 depending on the market. Not cheap. Now add a full staff of networking experts and admins to that. With a huge server farm you can cut staff requirements by more than 1/3. Over the course of a few years, that alone saves enough for the added cost of redundancy.

Plus, there is another growing trend of small and mid sized companies that do not want and can not afford an IT staff. Solution, thin client connections to an offsite, backup up, and redundant leased server. It will provide them with 10x more reliability than an in house IT staff.

enter the ASP. applicatin service provider. That market is growing strongly.

 

[polm]

I've supported large call centers as a Systems Administrator in control of, among other things, a pretty large terminal server farm. This, at the time, supported around 2000 thin clients.

Today I am a network analyst for the same company (well, bought and sold a few times over) with an even larger thin client setup, supporting around 10,000 users.

It's the ONLY way to go in an outfit that large. Anything else would be a freaking NIGHTMARE. This is strictly speaking from a support perspective. The cost alone is, as noted, well worth it.

 

[nweaver]

Originally posted by: aqt4u2lv
Dude, I'm talking about the software.



If you are logged into one server running word and someone is on another running something else it won't know about just because you have some magic switch.

SAN's, NAS, etc. All of the servers are writing to network storage, with access to it all. You save it on server1, but server 1 is on the same LUN (SAN "Hard Drive") as servers 2, 3, 4, 5 and 6.

I liked what someone said above, redundancy and failover. Throw in there fault tolerance and you have some planning and thought going into the multi million doller infrastructure. That is how they can move to this model, without fear of SPF of one or 2 servers.