Jump to content

Teaming

jathor

I have many options with my intel network card, what's the difference between link aggregation and adaptive load balancing? (when i select either it gives me 2gbps speeds). But which one is actually faster/better and what do they do?

 

thanks guys

Link to comment
Share on other sites

Link to post
Share on other sites

Link aggregation is the RAID of networking, numerous ethernet ports.

 

Load balancing is spreading a network's load across the board, balancing it out.

The projects never end in my line of work.

CPU: Dual Xeon E5-2650v2 || GPU: Dual Quadro K5000 || Motherboard: Asus Z9PE-D8 || RAM: 64GB Corsair Vengeance || Monitors: Dual LG 34UM95, NEC MultiSync EA244UHD || Storage: Dual Samsung 850 Pro 256GB in Raid 0, 6x WD Re 4TB in Raid 1 || Sound: Xonar Essense STX (Mainly for Troubleshooting and listening test) || PSU: Corsair Ax1500i

CPU: Core i7 5820k @ 4.7GHz || GPU: Dual Titan X || Motherboard: Asus X99 Deluxe || RAM: 32GB Crucial Ballistix Sport || Monitors: MX299Q, 29UB65, LG 34UM95 || Storage: Dual Samsung 850 EVO 1 TB in Raid 0, Samsung 850 EVO 250GB, 2TB Toshiba scratch disk, 3TB Seagate Barracuda || PSU: EVGA 1000w PS Platinum

Link to comment
Share on other sites

Link to post
Share on other sites

Both are technically the same thing 
Link aggregation will send half of the traffic on one link and the other half on the other (Use both link at the same time).
The Intel adaptive load balancing will use one link at its full capacity before starting to use the 2nd one.

 

 

Main: Intel i7-4770K, MSI Z87 MPower, 2x 4GB G.Skills 1866MHz CL9, 2x Gigabyte R9-290X, Fractal Design XL R2, 2x Samsung EVO 250GB Raid0 and 2x 1TB Seagate Raid0, Top power 1000W 80+ Bronze, Asus PCE-AC68

Server: AMD FX-8350, MSI 970a-G46, 2x 4GB Mushkin 1866Mhz CL9, MSI 8600GT, Fractal Design Define R4, Samsung EVO 250GB and 6X 2TB Seagate RAID 10, EVGA 850W Supernova G2 80+Gold, 2x Intel GBe NIC
Router: Pfsense 2.2, Intel Core 2 Quad Q6600, Acer crap motherboard, 4x 2GB Samsung 800Mhz CL4, Fractal Design Node 604, OCZ Agility 2 50GB, Thermaltake 400W 80+, 2x Intel GBe NIC

Link to comment
Share on other sites

Link to post
Share on other sites

Both are technically the same thing 

Link aggregation will send half of the traffic on one link and the other half on the other (Use both link at the same time).

The Intel adaptive load balancing will use one link at its full capacity before starting to use the 2nd one.

 

That isn't how load balancing works...

 

"In computingload balancing distributes workloads across multiple computing resources, such as computers, a computer clusternetwork links, central processing units or disk drives. Load balancing aims to optimize resource use, maximize throughput, minimize response time, and avoid overload of any single resource. Using multiple components with load balancing instead of a single component may increase reliability through redundancy. Load balancing usually involves dedicated software or hardware, such as a multilayer switch or a Domain Name System server process."

 

Source

The projects never end in my line of work.

CPU: Dual Xeon E5-2650v2 || GPU: Dual Quadro K5000 || Motherboard: Asus Z9PE-D8 || RAM: 64GB Corsair Vengeance || Monitors: Dual LG 34UM95, NEC MultiSync EA244UHD || Storage: Dual Samsung 850 Pro 256GB in Raid 0, 6x WD Re 4TB in Raid 1 || Sound: Xonar Essense STX (Mainly for Troubleshooting and listening test) || PSU: Corsair Ax1500i

CPU: Core i7 5820k @ 4.7GHz || GPU: Dual Titan X || Motherboard: Asus X99 Deluxe || RAM: 32GB Crucial Ballistix Sport || Monitors: MX299Q, 29UB65, LG 34UM95 || Storage: Dual Samsung 850 EVO 1 TB in Raid 0, Samsung 850 EVO 250GB, 2TB Toshiba scratch disk, 3TB Seagate Barracuda || PSU: EVGA 1000w PS Platinum

Link to comment
Share on other sites

Link to post
Share on other sites

That isn't how load balancing works...

 

"In computingload balancing distributes workloads across multiple computing resources, such as computers, a computer clusternetwork links, central processing units or disk drives. Load balancing aims to optimize resource use, maximize throughput, minimize response time, and avoid overload of any single resource. Using multiple components with load balancing instead of a single component may increase reliability through redundancy. Load balancing usually involves dedicated software or hardware, such as a multilayer switch or a Domain Name System server process."

 

Source

Well i guess you can't read because this is exactly what I wrote

BUT THE INTEL version of adaptative load balancing is using one link until full then switch over to the second link.

Also read the whole page of your own source ... you might need it :P

Main: Intel i7-4770K, MSI Z87 MPower, 2x 4GB G.Skills 1866MHz CL9, 2x Gigabyte R9-290X, Fractal Design XL R2, 2x Samsung EVO 250GB Raid0 and 2x 1TB Seagate Raid0, Top power 1000W 80+ Bronze, Asus PCE-AC68

Server: AMD FX-8350, MSI 970a-G46, 2x 4GB Mushkin 1866Mhz CL9, MSI 8600GT, Fractal Design Define R4, Samsung EVO 250GB and 6X 2TB Seagate RAID 10, EVGA 850W Supernova G2 80+Gold, 2x Intel GBe NIC
Router: Pfsense 2.2, Intel Core 2 Quad Q6600, Acer crap motherboard, 4x 2GB Samsung 800Mhz CL4, Fractal Design Node 604, OCZ Agility 2 50GB, Thermaltake 400W 80+, 2x Intel GBe NIC

Link to comment
Share on other sites

Link to post
Share on other sites

Well i guess you can't read because this is exactly what I wrote

BUT THE INTEL version of adaptative load balancing is using one link until full then switch over to the second link.

Also read the whole page of your own source ... you might need it :P

 

I suggest you read what I posted...

 

"maximize throughput, minimize response time, and avoid overload of any single resource"

 

They wouldn't max out one connection and then move on to another... That isn't balanced at all.

 

Aggregation is on one card, load balancing is over an entire network.

The projects never end in my line of work.

CPU: Dual Xeon E5-2650v2 || GPU: Dual Quadro K5000 || Motherboard: Asus Z9PE-D8 || RAM: 64GB Corsair Vengeance || Monitors: Dual LG 34UM95, NEC MultiSync EA244UHD || Storage: Dual Samsung 850 Pro 256GB in Raid 0, 6x WD Re 4TB in Raid 1 || Sound: Xonar Essense STX (Mainly for Troubleshooting and listening test) || PSU: Corsair Ax1500i

CPU: Core i7 5820k @ 4.7GHz || GPU: Dual Titan X || Motherboard: Asus X99 Deluxe || RAM: 32GB Crucial Ballistix Sport || Monitors: MX299Q, 29UB65, LG 34UM95 || Storage: Dual Samsung 850 EVO 1 TB in Raid 0, Samsung 850 EVO 250GB, 2TB Toshiba scratch disk, 3TB Seagate Barracuda || PSU: EVGA 1000w PS Platinum

Link to comment
Share on other sites

Link to post
Share on other sites

I suggest you read what I posted...

 

"maximize throughput, minimize response time, and avoid overload of any single resource"

 

They wouldn't max out one connection and then move on to another... That isn't balanced at all.

 

Aggregation is on one card, load balancing is over an entire network.

 

In computer networking, the term link aggregation applies to various methods of combining (aggregating) multiple network connections in parallel in order to increase throughput beyond what a single connection could sustain, and to provide redundancy in case one of the links should fail.

Regardless of the layer on which aggregation occurs, it balances the network load across all links. Most methods provide failover as well.

 
 
Yes you are right with the definition of load balancing BUT AGAIN INTEL VERSION OF ADAPTATIVE LOAD BALANCING 

INTEL VERSION

INTEL VERSION

I dont care how wikipedia defines load balancing its not how intel version does it.

Several carrier pay per LINK which means they will use one link at full capacity before switching to another one (BECAUSE COSTS)

 

Main: Intel i7-4770K, MSI Z87 MPower, 2x 4GB G.Skills 1866MHz CL9, 2x Gigabyte R9-290X, Fractal Design XL R2, 2x Samsung EVO 250GB Raid0 and 2x 1TB Seagate Raid0, Top power 1000W 80+ Bronze, Asus PCE-AC68

Server: AMD FX-8350, MSI 970a-G46, 2x 4GB Mushkin 1866Mhz CL9, MSI 8600GT, Fractal Design Define R4, Samsung EVO 250GB and 6X 2TB Seagate RAID 10, EVGA 850W Supernova G2 80+Gold, 2x Intel GBe NIC
Router: Pfsense 2.2, Intel Core 2 Quad Q6600, Acer crap motherboard, 4x 2GB Samsung 800Mhz CL4, Fractal Design Node 604, OCZ Agility 2 50GB, Thermaltake 400W 80+, 2x Intel GBe NIC

Link to comment
Share on other sites

Link to post
Share on other sites

 

In computer networking, the term link aggregation applies to various methods of combining (aggregating) multiple network connections in parallel in order to increase throughput beyond what a single connection could sustain, and to provide redundancy in case one of the links should fail.

Regardless of the layer on which aggregation occurs, it balances the network load across all links. Most methods provide failover as well.

 
 
Yes you are right with the definition of load balancing BUT AGAIN INTEL VERSION OF ADAPTATIVE LOAD BALANCING 

INTEL VERSION

INTEL VERSION

I dont care how wikipedia defines load balancing its not how intel version does it.

Several carrier pay per LINK which means they will use one link at full capacity before switching to another one (BECAUSE COSTS)

 

 

 

First off, calm down with the caps. It's unnecessary. He asked about the definitions, and while they are similar, they are for different scale deployments.

 

I understand he said Intel, I did read the post.

 

  • Adaptive Load Balancing (ALB)

    Offers increased network bandwidth by allowing transmission over 2-8 ports to multiple destination addresses, and incorporates Adapter Fault Tolerance. Only the primary adapter receives incoming traffic. Broadcasts/multicasts and non-routed protocols are only transmitted via the primary adapter in the team. The Intel® ANS software load balances transmissions, based on Destination Address, and can be used with any switch. Simultaneous transmission only occurs at multiple addresses. This mode can be connected to any switch.

That's straight off of Intel's site.

 

Link aggregation is using multiple ports for throughput more than redundancy.

Balancing makes transmission and receives go different routes.

 

It isn't the same thing though, sorry. 

 

And what "carrier" are you talking about that pay per link? You could have an internal server with absolutely no outside connection and still need balancing, especially moving large amounts of data across servers.

The projects never end in my line of work.

CPU: Dual Xeon E5-2650v2 || GPU: Dual Quadro K5000 || Motherboard: Asus Z9PE-D8 || RAM: 64GB Corsair Vengeance || Monitors: Dual LG 34UM95, NEC MultiSync EA244UHD || Storage: Dual Samsung 850 Pro 256GB in Raid 0, 6x WD Re 4TB in Raid 1 || Sound: Xonar Essense STX (Mainly for Troubleshooting and listening test) || PSU: Corsair Ax1500i

CPU: Core i7 5820k @ 4.7GHz || GPU: Dual Titan X || Motherboard: Asus X99 Deluxe || RAM: 32GB Crucial Ballistix Sport || Monitors: MX299Q, 29UB65, LG 34UM95 || Storage: Dual Samsung 850 EVO 1 TB in Raid 0, Samsung 850 EVO 250GB, 2TB Toshiba scratch disk, 3TB Seagate Barracuda || PSU: EVGA 1000w PS Platinum

Link to comment
Share on other sites

Link to post
Share on other sites

First off, calm down with the caps. It's unnecessary. He asked about the definitions, and while they are similar, they are for different scale deployments.

 

I understand he said Intel, I did read the post.

 

  • Adaptive Load Balancing (ALB)

    Offers increased network bandwidth by allowing transmission over 2-8 ports to multiple destination addresses, and incorporates Adapter Fault Tolerance. Only the primary adapter receives incoming traffic. Broadcasts/multicasts and non-routed protocols are only transmitted via the primary adapter in the team. The Intel® ANS software load balances transmissions, based on Destination Address, and can be used with any switch. Simultaneous transmission only occurs at multiple addresses. This mode can be connected to any switch.

That's straight off of Intel's site.

 

Link aggregation is using multiple ports for throughput more than redundancy.

Balancing makes transmission and receives go different routes.

 

It isn't the same thing though, sorry. 

 

And what "carrier" are you talking about that pay per link? You could have an internal server with absolutely no outside connection and still need balancing, especially moving large amounts of data across servers.

Thank you for confirming exactly what I said since the begining !

Main: Intel i7-4770K, MSI Z87 MPower, 2x 4GB G.Skills 1866MHz CL9, 2x Gigabyte R9-290X, Fractal Design XL R2, 2x Samsung EVO 250GB Raid0 and 2x 1TB Seagate Raid0, Top power 1000W 80+ Bronze, Asus PCE-AC68

Server: AMD FX-8350, MSI 970a-G46, 2x 4GB Mushkin 1866Mhz CL9, MSI 8600GT, Fractal Design Define R4, Samsung EVO 250GB and 6X 2TB Seagate RAID 10, EVGA 850W Supernova G2 80+Gold, 2x Intel GBe NIC
Router: Pfsense 2.2, Intel Core 2 Quad Q6600, Acer crap motherboard, 4x 2GB Samsung 800Mhz CL4, Fractal Design Node 604, OCZ Agility 2 50GB, Thermaltake 400W 80+, 2x Intel GBe NIC

Link to comment
Share on other sites

Link to post
Share on other sites

Thank you for confirming exactly what I said since the begining !

 

Similar is not the same. One is for safety, the other is for throughput. You said they are the same ("they are technically the same thing"), and they aren't.

 

@jathor for personal use, neither one is necessary. 

The projects never end in my line of work.

CPU: Dual Xeon E5-2650v2 || GPU: Dual Quadro K5000 || Motherboard: Asus Z9PE-D8 || RAM: 64GB Corsair Vengeance || Monitors: Dual LG 34UM95, NEC MultiSync EA244UHD || Storage: Dual Samsung 850 Pro 256GB in Raid 0, 6x WD Re 4TB in Raid 1 || Sound: Xonar Essense STX (Mainly for Troubleshooting and listening test) || PSU: Corsair Ax1500i

CPU: Core i7 5820k @ 4.7GHz || GPU: Dual Titan X || Motherboard: Asus X99 Deluxe || RAM: 32GB Crucial Ballistix Sport || Monitors: MX299Q, 29UB65, LG 34UM95 || Storage: Dual Samsung 850 EVO 1 TB in Raid 0, Samsung 850 EVO 250GB, 2TB Toshiba scratch disk, 3TB Seagate Barracuda || PSU: EVGA 1000w PS Platinum

Link to comment
Share on other sites

Link to post
Share on other sites

Similar is not the same. One is for safety, the other is for throughput. You said they are the same ("they are technically the same thing"), and they aren't.

 

@jathor for personal use, neither one is necessary. 

 

basically what I would like is to get the fastest speed from the (home/media) server in order to watch multiple hd streams of movies and music from it (multiple people living with me who have a thirst for movies/shows/music) and transfer files to and from it without affecting (or not affecting much) the speed and/or quality of the stream.

Link to comment
Share on other sites

Link to post
Share on other sites

basically what I would like is to get the fastest speed from the (home/media) server in order to watch multiple hd streams of movies and music from it (multiple people living with me who have a thirst for movies/shows/music) and transfer files to and from it without affecting (or not affecting much) the speed and/or quality of the stream.

 

It still likely isn't needed, can your router monitor your LAN bandwidth? I'd check that and see if you are maxing 1Gbps before spending any money/time on upgrading.

 

I have LAG going to my server and out to my workshop simply because I didn't already have the hardware and I only ended up spending $30 more to get switches with LAG capability. I never would have messed with it if I already had switches in place.

 

I could be wrong, but I believe the "typical" HD stream is around 7MBps, so even if you had 5 people watching different movies at the same time you would have 90MBps of overhead which is only going to be a noticeable bottleneck if you do a lot of sequential transfers, use SSDs on both ends, or have a nice RAID setup.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×