Lecture 135: DX Link Aggregation Groups (LAGs)
- 课程：[NEW] AWS Certified Advanced Networking Specialty 2026
- 视频时长：07:37
- 来源：Udemy Lecture 27898752
- Slide 数量：26
- 画面切换：00:21、00:38、00:58、01:23、01:35、01:42、01:45、02:01、03:01、03:11、03:23、03:32、03:51、03:56、04:02、04:33、04:54、05:00、05:28、05:44、05:58、06:15、06:37、07:02、07:12
扫描说明
逐条检查了 Udemy 页面中的 111 条 Transcript cue，并扫描了全长约 7:37 的视频。依据 LAG 约束逐条渐显、1/10/100 Gbps 聚合示例、双 LAG 高可用拓扑、运行状态阈值与 MACsec 密钥规则，共划分为 26 个小节。正式截图阶段遇到浏览器会话超时后重新连接，并改为单张抓取；26 张截图均在视频播放、控件隐藏、Transcript 与课程侧栏关闭后按真实媒体边界截取，逐张确认无黑边、无右侧或底部裁切、无播放按钮或覆盖层，并统一输出为 1364×770。
Transcript 来自 Udemy 页面自带的英文自动字幕，没有使用 ASR。下面保留页面原文，包括其中的语法或识别错误。
1. 链路聚合组（LAG）概览
时间范围：00:00–00:21
总结 / Summary
中文
链路聚合组（Link Aggregation Group，LAG）把多条 AWS Direct Connect（DX）物理连接组合成一条逻辑连接。
English
A Link Aggregation Group combines multiple AWS Direct Connect physical connections into one logical connection.
Transcript
00:00 Hi.
00:01 So in this lecture let's understand another feature of AWS Direct Connect that is link aggregation groups
00:07 or called as lags.
00:09 Now, as the name suggests, it aggregates multiple links.
00:13 That means multiple direct connect connections.
00:16 So let's see what's the purpose and why should you use the lags.
2. 聚合连接提升带宽
时间范围：00:21–00:38
总结 / Summary
中文
通过聚合多条 Direct Connect（DX）连接，LAG 提供汇总带宽（Aggregate Bandwidth）并简化逻辑管理。
English
By aggregating multiple Direct Connect links, a LAG provides combined bandwidth and simpler logical management.
Transcript
00:21 So the basic reason for using the lags is to aggregate multiple DX connections so that if, say you
00:28 have four connections with one GB s bandwidth, you can aggregate that into a single pipe of four GB
00:34 s connection.
00:36 So using the lags you get increased bandwidth.
3. 聚合同时提供故障转移
时间范围：00:38–00:58
总结 / Summary
中文
当某一成员连接故障时，其余成员仍可承载流量，因此 LAG 同时提供故障转移（Failover）能力。
English
If one member connection fails, the remaining members continue carrying traffic, so a LAG also provides failover.
Transcript
00:38 And also you get a failover failover in the sense that if there are four connections, even if one connection
00:44 fails, then also your lags still operates.
00:48 So you will get maybe three GB s bandwidth, but your connection will work so it consolidate all your
00:54 connections into a single logical connection.
4. LAG 使用 LACP
时间范围：00:58–01:23
总结 / Summary
中文
LAG 使用链路聚合控制协议（Link Aggregation Control Protocol，LACP）协商并维护聚合成员。
English
A LAG uses the Link Aggregation Control Protocol to negotiate and maintain its member links.
Transcript
00:58 Now for lag it uses a protocol called Link Aggregation Control Protocol Lacp.
01:05 Now there are few prerequisites or kind of considerations when you can create the lags.
01:10 So now let's understand that.
01:12 So for creating the lags your connection should be dedicated connections.
01:17 And the bandwidth of those connection could be one GB ten GB or 100 GB s.
5. 仅支持相同速率的 Dedicated Connections
时间范围：01:23–01:35
总结 / Summary
中文
所有成员必须是专用连接（Dedicated Connections），并使用相同的 1、10 或 100 Gbps 端口速率；托管连接（Hosted Connections）不能加入 LAG。
English
All members must be dedicated connections at the same supported 1, 10, or 100 Gbps port speed; hosted connections cannot join a LAG.
Transcript
01:23 Which means if you have hosted DCS connection then you cannot create the lag on that.
01:28 Now it can aggregate up to maximum four connections only and those connections will be in active mode.
6. 最多聚合四条连接
时间范围：01:35–01:42
总结 / Summary
中文
一般情况下一个 LAG 最多聚合四条连接，并以双活（Active-Active）方式运行。
English
A LAG generally aggregates up to four connections operating in active-active mode.
Transcript
01:35 Now you can either have the existing connections added to the lag group, or you can create a new connections
7. 可使用现有或新建连接
时间范围：01:42–01:45
总结 / Summary
中文
创建 LAG 时既可迁移现有 Direct Connect（DX）连接，也可新建成员连接。
English
A LAG can be built from existing Direct Connect links or from newly provisioned member connections.
Transcript
01:42 and create a lag out of those connections.
8. 所有成员带宽必须一致
时间范围：01:45–02:01
总结 / Summary
中文
LAG 内不能混合不同速率的连接；所有成员必须具有相同带宽（Bandwidth）。
English
A LAG cannot mix connection speeds; every member must have the same bandwidth.
Transcript
01:45 Now all connections in the lags must have the same bandwidth, so you cannot mix like one GB s connection
01:51 and ten GB s connection.
01:53 All should be one GB, all should be ten GB or 100 GB s, and all these connections should be terminating
9. 所有成员终止于同一 DX 端点
时间范围：02:01–03:01
总结 / Summary
中文
所有成员必须终止于同一 AWS Direct Connect 端点（DX Endpoint）和同一位置，不能跨位置构成单个 LAG。
English
All members must terminate at the same AWS Direct Connect endpoint and location; a single LAG cannot span locations.
Transcript
02:01 on the same DCS location.
02:04 So this is an interesting because if you already have one connection and now you ask AWS to have another
02:10 connection, typically for better fault tolerance, AWS will allocate the DCS port on different device,
02:18 but for the lag you need to terminate all these connections on the same device.
02:22 So while requesting the connection, you can ask AWS that provide me this connection on which I have
02:30 existing connection.
02:32 So this is an important consideration while creating the lag.
02:36 And finally, once you have the lag, then you can create your public virtual interface, private virtual
02:42 interface, or transit virtual interface.
02:44 So you basically treat your lag as your DCS connection.
02:49 So logically there is then no difference between the connection and the lag.
02:53 Right.
02:53 With that.
02:54 Now let's look at quickly few architectures and few more things about the lags that you should know
02:59 for your exam.
10. 1 Gbps LAG 示例拓扑
时间范围：03:01–03:11
总结 / Summary
中文
示例从四条独立的 1 Gbps 专用连接（Dedicated Connections）开始，连接同一 DX 设备与客户路由器。
English
The example starts with four independent 1 Gbps dedicated connections between the same DX device and customer router.
Transcript
03:01 So this is the architecture that we have been seeing for AWS TX connections.
03:06 But now what we are saying that instead of one connection from your corporate data center, you could
11. 四条 1 Gbps 成员链路
时间范围：03:11–03:23
总结 / Summary
中文
四条成员链路都为 1 Gbps，满足同速率和同端点要求。
English
All four member links run at 1 Gbps, satisfying the equal-speed and same-endpoint requirements.
Transcript
03:11 have multiple connections of say one GB s, and then instead of having four connections and getting
03:18 one GB s on every connection, you can create a lag out of this four connection.
12. 1 Gbps 成员形成 4 Gbps LAG
时间范围：03:23–03:32
总结 / Summary
中文
LACP 将四条 1 Gbps 链路组合成一个逻辑 LAG，总有效带宽（Effective Bandwidth）为 4 Gbps。
English
LACP combines the four 1 Gbps links into one logical LAG with 4 Gbps of effective bandwidth.
Transcript
03:23 So you get effective bandwidth of four GB s.
03:27 Right.
03:27 So that's basically a lag.
03:29 Now you could also have ten GB s connection.
13. 10 Gbps 成员形成 40 Gbps LAG
时间范围：03:32–03:51
总结 / Summary
中文
四条 10 Gbps 专用连接（Dedicated Connections）可组成 40 Gbps 的聚合链路。
English
Four 10 Gbps dedicated connections can form a 40 Gbps aggregated link.
Transcript
03:32 And in that case your effective bandwidth will be 40 GB s.
03:37 But as you go to the 100 GB s then you cannot have 400 GB s as your aggregated bandwidth.
03:44 As of today, AWS only allows to have 200 GB s connections to be a part of the lag.
14. 100 Gbps LAG 示例
时间范围：03:51–03:56
总结 / Summary
中文
100 Gbps 连接也能用于链路聚合组（LAG），但成员数量限制不同。
English
A Link Aggregation Group can also use 100 Gbps links, but its member-count limit differs.
Transcript
03:51 So the effective bandwidth you can get up to maybe 200 GB s.
15. 100 Gbps LAG 最多两条连接
时间范围：03:56–04:02
总结 / Summary
中文
100 Gbps 连接的 LAG 最多包含两条成员链路，因此最高有效带宽（Effective Bandwidth）为 200 Gbps。
English
A 100 Gbps LAG supports at most two member links, yielding up to 200 Gbps of effective bandwidth.
Transcript
03:56 So while using the 100 GB s, there is this difference that you should understand.
04:01 Okay.
16. 使用两个 LAG 构建高可用
时间范围：04:02–04:33
总结 / Summary
中文
两个独立链路聚合组（LAGs）连接不同设备，可降低单个 LAG、客户路由器或 DX 设备故障的影响，形成更高层级的高可用（High Availability）。
English
Two independent LAGs using separate devices reduce the impact of a LAG, customer-router, or DX-device failure and provide higher-level availability.
Transcript
04:02 With that.
04:02 Now let's also talk about the high availability of the lags because they are like the direct connect
04:09 connections.
04:09 After you create the lag, you can have similar high availability architecture that you had with the
04:15 DCS connections.
04:17 So you can have two different lags.
04:19 And then if one lags fail, you can have another lag to take over that connection.
04:25 So this is also possible.
04:27 Now for your exam you also need to know something called operational status of the lag.
04:32 Now what that means.
17. LAG 运行状态概览
时间范围：04:33–04:54
总结 / Summary
中文
LAG 的运行状态（Operational Status）取决于成员连接是否满足配置的最小存活数量。
English
A LAG's operational status depends on whether its member connections meet the configured minimum active count.
Transcript
04:33 So generally lag has multiple connections.
04:36 And you can define that.
04:38 If one of the connection fails then should my lag also show the status as failed or my lag should keep
04:45 operating.
04:46 So that's called operational status of the lag.
04:50 So as I said, all the connections in the lag are in active mode all the time.
18. 成员以 Active-Active 运行
时间范围：04:54–05:00
总结 / Summary
中文
LAG 中的所有成员连接都以双活（Active-Active）模式转发流量，而不是固定主备。
English
All member connections in a LAG forward traffic in active-active mode rather than as fixed primary and backup links.
Transcript
04:54 But you can define the operational parameters, which defines the minimum number of connections that
19. 最小运行连接数属性
时间范围：05:00–05:28
总结 / Summary
中文
最小运行连接数（Minimum Operational Connections）属性定义 LAG 保持可运行所需的最少活动成员；默认值为 0。
English
The minimum operational connections attribute defines how many active members are required for the LAG to remain operational; its default is zero.
Transcript
05:00 should be active for lag to be active.
05:03 So default is zero.
05:04 Which means if you have four connections in the lag and say three connections goes down, still your
05:10 lag is active, right?
05:11 But many a times what you want is that you want to make sure that at any moment of time, you need at
05:18 least three GB of bandwidth.
05:20 If bandwidth falls below that, you want to show the status of lag as failure.
05:25 So let's take an example that you have four connections.
20. 阈值 1：至少一条连接存活
时间范围：05:28–05:44
总结 / Summary
中文
四成员 LAG 的阈值设为 1 时，只要至少一条连接仍可用，LAG 就保持运行。
English
With a threshold of one on a four-member LAG, the LAG remains operational as long as at least one connection is available.
Transcript
05:28 And then you say operational connection attribute value is one, which means as long as at least one
05:34 connection is up, your lag is up.
05:37 But if you define this value as three, that means your lag will be non operational if two or more connections
21. 阈值 3：两条故障即停用
时间范围：05:44–05:58
总结 / Summary
中文
阈值设为 3 时，需要至少三条成员连接存活；若两条或更多连接故障，LAG 变为不可运行（Non-Operational）。
English
With a threshold of three, at least three members must stay active; the LAG becomes non-operational if two or more links fail.
Transcript
05:44 are down.
05:46 So in this way you can control the behavior of the lag.
05:50 So what's the purpose of this attributes.
05:52 So typically it makes sure that you are not over utilizing the active lag connection in case of failures
22. 运行阈值防止剩余链路过载
时间范围：05:58–06:15
总结 / Summary
中文
设置合理阈值可在容量下降过多时停用 LAG，避免剩余活动连接过载（Over-Utilization）。
English
A suitable threshold disables the LAG after excessive capacity loss, preventing over-utilization of the remaining active links.
Transcript
05:58 of the underlying connection.
06:00 So you want to show the status as failure if multiple connections also fails.
06:05 So that's the lag operational status.
06:09 Okay, just one more thing that you should know for the lag is that you can enable the macsec with the
23. LAG 支持 MACsec
时间范围：06:15–06:37
总结 / Summary
中文
链路聚合组（LAG）可以启用媒体访问控制安全（MAC Security，MACsec）来保护二层链路。
English
A Link Aggregation Group can enable MAC Security to protect its Layer 2 links.
Transcript
06:15 lag.
06:15 Macsec is max security, which works at layer two and used to protect your data flowing through the
06:21 DCS connections.
06:23 And we have talked about Macsec with respect to the DCS connections.
06:27 So same macsec can be applied for even the lag.
06:31 So when you enable the Macsec, it uses the Macsec secret key which is generated using the connection
24. MACsec 使用 CKN 与 CAK
时间范围：06:37–07:02
总结 / Summary
中文
MACsec 密钥材料通过连接密钥名称（Connection Key Name，CKN）与连接关联密钥（Connectivity Association Key，CAK）生成和标识。
English
MACsec key material is generated and identified using the Connection Key Name and Connectivity Association Key.
Transcript
06:37 key name CDN and connectivity association key.
06:40 That is c a k.
06:42 Now only the difference is that if you use existing connection to create a lag, and on that connection,
06:50 if you have the macsec already enabled, then that macsec key is dis associated with the connection
06:57 first, and the lag key instead associated with that connection.
25. 现有连接加入 LAG 后改用 LAG 密钥
时间范围：07:02–07:12
总结 / Summary
中文
启用 MACsec 的现有连接加入 LAG 后，会解除原连接级 MACsec 密钥关联，转而使用 LAG 级密钥。
English
When an existing MACsec-enabled connection joins a LAG, its connection-level MACsec key is disassociated and replaced by the LAG-level key.
Transcript
07:02 So this is just one thing that you should be knowing.
07:05 So your existing connection do not carry the same key as they become part of the lag.
26. 所有成员共享 LAG MACsec 密钥
时间范围：07:12–07:37
总结 / Summary
中文
LAG 的 MACsec 密钥（MACsec Key）会关联到所有成员连接，使整个聚合组使用一致的安全关联。
English
The LAG MACsec key is associated with every member connection so the entire aggregation uses a consistent security association.
Transcript
07:12 And which means all the members in the lag will use the same key as the lag Macsec key.
07:19 Right?
07:20 So that's another thing, right?
07:22 So that's it about the lag.
07:24 There isn't much to discuss here, but I hope you got the concept and you now know what the lag is and
07:31 why and where it is used.
07:33 So thanks for watching and stay tuned.
整讲总结 / Overall Summary
中文
链路聚合组（Link Aggregation Group，LAG）利用 LACP 把多条同速率、同端点的 Dedicated Direct Connect Connections 组合成单一逻辑链路，从而增加带宽并提供成员级故障转移。1 或 10 Gbps LAG 最多四条成员，100 Gbps LAG 最多两条；所有成员以 Active-Active 运行。最小运行连接数（Minimum Operational Connections）可在剩余容量不足时让 LAG 停止工作，避免链路过载。LAG 还支持 MACsec，并让所有成员共享 LAG 级 CKN/CAK 密钥关联。
English
A Link Aggregation Group uses LACP to combine dedicated Direct Connect connections with equal speed and a common endpoint into one logical link, increasing bandwidth and providing member-level failover. A 1 or 10 Gbps LAG supports up to four members, while a 100 Gbps LAG supports up to two; all members operate active-active. The minimum operational connections setting can take the LAG out of service when remaining capacity is insufficient, preventing overload. LAGs also support MACsec, with all members sharing the LAG-level CKN/CAK association.
返回顶部 ↑