Video: CARA MEMILIH QUEUE TYPES MIKROTIK – QOS [ENG SUB]
Hi guys See you again at the Mikrotik tutorial and in this tutorial I will tell you about the Queue Types if we make a queue usually we will be told to choose several queue types what are the differences from some of the queue types and when we use certain queues see this video but beforehand for you who haven't subscribed don't forget to click Subscribe and also click the bell so you will get a notification when we uploaded the latest video in this video we will talk about Queue Types so this is one of a series of videos about QoS and no less important when setting up QoS is to choose Queue Types because each time we create a new queue we must determine what queue types to use if we look at the simple queue menu on the simple queue menu there will be a queue types tab default setting is default-small later we will discuss that queue types are not only default-small because there are more standard queue types on the queue tree menu there are also queue types settings we see in the queue tree menu there is also a queue types selection when we create a queue tree there will be a choice for queue types the cloud settings will be default-small what are the queue types on Mikrotik? before we have to understand that for the QoS process there are 2 things that work the first is the Scheduler and Shaper Shaper is a method for limiting so who limits bandwidth for example 1MB, 10MB, 20MB is Shaper can be seen in the diagram that Shaper has the duty to protect Bandwidth then the other one is called the Scheduler Scheduler is a method of how we queue incoming data packets if all packet data is entered far below the limit actually there is no problem because all packet data will just pass but if Packet data passes greater than the Bandwidth limit then the Router will do the queue and the Router will do Buffering how is this buffer done? then what order sequence? this is a function of the Scheduler so in the QoS process there is a Shaper and there is a Scheduler the diagram will look different if the shaper process will immediately cut then the scheduler will queue up for Mikrotik QoS the two processes work together so we will very rarely meet both processes run separately both processes will work we will see these queue types working if we talk about shaper there will be two things that will be done by the Router the first is HTB or Hierarchial Token Bucket this is the mechanism by which we cut or limit against Bandwidth that will pass then there is also PCQ where we can divide Bandwidth evenly for some traffic that has been chosen by the Router and PCQ is included in the Shaper that will close then for the Scheduler there are several types the first we know as FIFO or First In First Out then the second is RED Random Early Detect and the third is SFQ Stochastic Fairness Queuing and also PCQ (Per Connection Queue) some of the types of Scheduler that we will choose when making a queue we will discuss one by one first from FIFO FIFO is the simplest Scheduler that is on the Router OS therefore this scheduler only requires the smallest resource for the same traffic for FIFO it is divided into 3 types which we can use first is PFIFO that works based on the number of Packet data the second is BFIFO who works based on large data Bytes or based on the large size of traffic and the third is MQ-FIFO or Multi Queue FIFO MQ-FIFO can only be used for Routers that have a Multi Core Processor because it will make several Queues that are compatible with the Multi Core Processor so choose PFIFO or BFIFO? if we need a very simple scheduler we can use PFIFO because the Router does not need to calculate the size of the Bytes that will be passed The router will only count how many packets will be missed how many packets will be encrypted so if we use a Router with a low CPU and Memory we can use PFIFO because this scheduler needs it The smallest resource if we see the Winbox menu when choosing Queue Type there will only be one choice of Queue Size parameters if we choose PFIFO we must determine how many packets will be queued whereas if we choose BFIFO we must determine how big the Byte will be queued the initial setting is default-small is 10 Packets to be queued meaning that there are already 10 data packets which queues in the buffer then the packet after that will be dropped so for transmissions that don't recognize connections like UDP then the data will be loss but for traffic like TCP then the server will resend the data packet this can cause a delay that increases if the buffer is full and we still have Ping traffic and we don't distinguish this Ping traffic on different queues then the Ping delay will also increase because the packet is buffered the second is the possibility of the packet being dropped so it must be resent for Queue Size or buffer size if we increase the buffer size then the Router will automatically use More resources so if we do limitation for the same bandwidth for the same number of packets if the queue size or number of buffers 10 or 100 the 100 will certainly require CPU load and also greater RAM so our RAM capacity will quickly be used up compared to if we use a queue size or small amount of buffer then for the BFIFO Scheduler if we use BFIFO the parameter is the amount of packet per Bytes the way it works is almost the same as PFIFO BFIFO will queue based on the size of the data packet not based on the number of packet data so based on the size of the data packet that enters the queue size or buffer for the second scheduler is RED or Random Early Detect from the name we can know that This scheduler will work randomly so there is a possibility of data packets will be dropped or forwarded there are several parameters that are used to determine which data packet will be forwarded and that will be dropped the first is the minimum threshold if the average data packet passes below the minimum threshold then the data packet will be forwarded or will enter into the queue if the average data packet passes above the maximum threshold then the data packet will be dropped if below will continue if the above will be dropped in the middle The router will perform calculations to determine which packet will be forwarded and which will be dropped so this method is also called the guessing method so it's called Random so the Router will do a random process which one will continue and which will be dropped so what we need to know is what is the minimum threshold and what is the maximum threshold if we look at the configuration on the router here is clear there is a queue size or how many data packets then the minimum threshold and maximum threshold then there is also Burst Burst here is not much different from the discussion in the previous video but this is a special Burst for RED RED is good for heavy traffic if our customers buy 10Mb of bandwidth but its use is always above 10Mbps it might be suitable to apply this RED Queue method but but this RED is only good for TCP traffic because because the server can reset the data if there is a drop but for UDP packets to run Online Games is not suitable if applying RED because later The packet will be considered by the router to meet Queue so it will drop earlier than the other connections so RED is good for dense networks but for TCP The third scheduler is SFQ or Queuing Stochastic Fairness this is a scheduler mechanism that can balance or balance traffic based on connection how does SFQ balance this? SFQ will make Sub-Queue a number of 1024 Sub-Queues how to separate it into Sub-Queue? SFQ will use a classifier like on PCQ but the classifier on the SFQ cannot be changed based on Source address, source port, destination address, and destination port so SFQ will immediately share the connection to be 1024 Sub-Queue where 1 Sub-Queue maximum size is 128 packets so after there are several Sub-Queues SFQ will do round robin and also hashing will choose which Sub-Queue who may pass first so if we have many connections then SFQ will balance these connections the drawback is if we use a downloader application or use a torrent and these applications will open a lot of connections then the Sub-Queue for this queue might run out so that to skip other data traffic or more importantly it won't get past this is often the case with SFQ if we see the settings in Winbox there are only two parameters namely Perturb and Alot Perturb is the time needed by the Router to refresh hashing the initial setting will be written 5 seconds every 5 seconds the router will refresh hasing so that it will create a new Sub-Queue every 5 seconds there is one more parameter, Alot which is the amount of data that is transmitted here it is written in units of Bytes like a buffer with Bytes instead of a packet SFQ is widely used if we make many connections but we want to connect to it not interfere with each other how about all connections can get Bandwidth the problem is if our connection is too much, there will be several packets that do not get Sub-Queue so that it will feel slow for the application then there is another scheduler namely PCQ and this scheduler is very special because can function as a scheduler and also as a shaper so this is the only QoS process method that combines sheduler and shaper PCQ or short for Per Connection Queue from the name we can know that PCQ works based on connection this method is an improvement from SFQ the most felt is the first is there is no limitation on the number of Sub-Queues on SFQ we know that the maximum Sub-Queue is 1024 whereas in PCQ there is no limit to the Sub-Queue but at the limit based on the use of RAM and also the CPU load so hardware capabilities have an effect on the use of PCQ The second for the classifier if on the SFQ, the classifier cannot be replaced by default it will use 4 classifiers if we use PCQ we can choose which one to use classifier so that we can also use PCQ for example to divide Bandwidth along with its portion for all IPs we put aside the Port first we will divide based on IP IP source or IP destination then we will use the source address classifier or destination address then the Router will divide the connection speed based on the number of IP detected so if it's based on source address then the router will automatically divide based on a number of source addresses and also applies to destination addresses if we want to make per connection like SFQ we need to use 4 available classifiers then PCQ will work similar to SFQ PCQ is usually used for Automatically bandwidth Bandwidth per IP for example we have 100 clients we don't need to make rules one by one because we can use PCQ because we can specify the classifier then we can make a different type in one PCQ one for upload one for download we usually use uploads classifier source address because we distinguish connections based Origin IP address while for download traffic we use PCQ with the classifier destination address because we will divide the connection based on the destination IP so if data is entered we will divide based on destination address then for the data out we divide based on source address PCQ will automatically share Bandwidth based on a number of our clients therefore it's on a simple queue there are two queue types options first to download and one to upload don't turn it upside down to determine upload and download classifier while in the queue tree usually we will make a rule to download and one rule to upload therefore we can only choose one queue type used to download or to upload one of the most favorite things from PCQ and because that's what people use a lot is about PCQ rate so on this PCQ rate for example we have large bandwidth for example, we empty the PCQ rate automatically PCQ will share Bandwidth fairly for example we have a 10 Mbps Bandwidth and we have 10 customers it will automatically divided into 1 Mbps each if there are only two clients it will be divided into 5 Mbps respectively for example, we will protect you from the customer using bandwidth up to 5 Mbps even though Bandwidth is still left and unused we only make a maximum of 2 Mbps then we can protect the PCQ rate by 2 Mbps so that the maximum customer can only use Bandwidth up to 2 Mbps and if there are many clients and must be divided evenly then from the available bandwidth will be distributed evenly to all clients for example we have a 10 Mbps Bandwidth and the router detects there are 10 IP clients it will be divided into 1 Mbps all if the router detects there are 20 IP clients it will be divided into 512Kbps per IP client but if the router only detects there are only 1, 2 or 3 Ip clients and before we have made a PCQ rate of 2 Mbpas then each client will get Only 2 Mbps bandwidth can't use based on all available bandwidth sharing because we made the previous PCQ rate that's why PCQ is referred to as scheduler and also as a shaper because it can queue and also make limitations for PCQ there are several parameters we need to look at too the first one already has a PCQ rate which is used to determine maximum limit per client maximum limit per Sub-Queue for example there is only one Sub-Queue then the maximum bandwidth can only be reaching as big as PCQ rate even though the client has not reached the max limit if the client has reached the PCQ rate limit then the client will only get the Bandwidth access for the PCQ rate there are two more parameters that actually exist on PCQ and are rarely used but actually it's also important to note namely limit parameters and total limit parameters with a Packet count what is the Limit packet and the Total limit Limit is the buffer size for each Sub-Queue for example here the initial setting is 50 Packet Limit then for Total Limit is the total number of buffers for the queue we must pay attention to we must be able to calculate how many Sub-Queues there will be whether 10 Sub-Queue, whether 20 or 100 for example there will be 100 Sub-Queues and we set the limit per Sub-Queue to 50 then the total limit is multiplication from the limit and the number of Sub-Queues so if the limit is 50 then Sub-Queue is 100 then the total limit is 50 x 100 = 5000 we are often wrong here because we don't make changes we only use settings existing queue types which will cause the buffer to be small if we have a lot of clients then the limit is big so at times with the number of Sub-Queues the total limit is not as big as the calculation so we have to calculate it correctly must be determined in advance how many limits and how many Sub-Queues the results of these calculations which becomes the total limit such is the explanation of queue types which is owned by Mikrotik RouterOS basically we have to know the difference between the shaper and also the scheduler and when we use these queue types the simplest we can use FIFO for those who are more advanced or who have solid traffic we can use RED or we just want to balance traffic which is in the queue we can use SFQ and if we want to balance and share the traffic of a number of clients that we have or number of destinations we can use PCQ so when we make the queue can be a simple queue or queue tree we must know will use which queue types don't let us choose wrongly fit to the hardware and also typical of our clients adjust also with the network that we have so we can choose queue types the most appropriate for our network don't forget that we will still make it some more videos in this QoS series because some of these videos are still basic concepts watch some of our next videos in the QoS series for you who haven't subscribed Don't forget to subscribe and also click the bell button so you can get notifications or you have experience and would like to comment on these Queue Types can comment in the comments column below don't forget Share so that more people will know about the Queue Types See you in the next video.