{"id":25245,"date":"2026-11-17T12:04:56","date_gmt":"2026-11-17T12:04:56","guid":{"rendered":"https:\/\/atalnetworks.com\/?p=25245"},"modified":"2026-11-17T12:04:56","modified_gmt":"2026-11-17T12:04:56","slug":"why-is-my-10gbps-server-slow","status":"publish","type":"post","link":"https:\/\/atalnetworks.com\/ar\/why-is-my-10gbps-server-slow\/","title":{"rendered":"Why Is My 10Gbps Server Slow? A Step-by-Step Troubleshooting Guide"},"content":{"rendered":"<p><span style=\"font-weight: 400;\">A 10Gbps server can be slow because the client, storage, CPU, application, or network path cannot sustain the requested rate. Start by checking speed units and link speed. Then compare controlled tests in both directions while measuring CPU use, interface errors, and TCP behavior.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A fast server port is only one part of a transfer. The receiving device, remote network, and software must also keep up.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">We recommend working from the symptom to a controlled test. Each result should narrow the search before you change a setting. This guide uses Linux examples and explains what the results can and cannot prove.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Start Here: Match Your Symptom to the Next Test<\/span><\/h2>\n<table>\n<thead>\n<tr>\n<th><b>What you see<\/b><\/th>\n<th><b>Possible causes<\/b><\/th>\n<th><b>What to check next<\/b><\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><span style=\"font-weight: 400;\">Transfers stay around 110\u2013120 MB\/s<\/span><\/td>\n<td><span style=\"font-weight: 400;\">A 1Gbps segment, storage limit, or configured cap<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Check both endpoints, link rates, and service limits<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">One connection is slow, but several are faster<\/span><\/td>\n<td><span style=\"font-weight: 400;\">A per-flow limit, CPU processing, TCP behavior, or different flow paths<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Compare one and four streams while watching individual CPU cores<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">A network test is fast, but file copies are slow<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Storage, encryption, application behavior, or different traffic policies<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Compare disk activity and application CPU use during the copy<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Only one destination is slow<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Remote endpoint limits, routing, congestion, or filtering<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Repeat the same test against another approved endpoint<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Uploads and downloads differ<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Sender or receiver limits, directional policies, or different paths<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Reverse the transfer and monitor both hosts<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Performance drops at busy times<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Competing workloads, shared capacity, or queue buildup<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Compare host and network metrics at matching timestamps<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">One CPU core stays busy<\/span><\/td>\n<td><span style=\"font-weight: 400;\">A worker thread, packet-processing queue, or encryption limit<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Check per-core use and the process handling traffic<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Small requests work, but large transfers stall<\/span><\/td>\n<td><span style=\"font-weight: 400;\">MTU problems, filtering, or application issues<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Check path MTU and whether the problem follows packet size<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Latency rises during a transfer<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Queues filling on a host or network device<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Compare latency before, during, and after the load<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">An intermediate MTR hop shows loss<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Limited diagnostic replies or real forwarding trouble<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Check later hops and compare with application symptoms<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><span style=\"font-weight: 400;\">These are starting points. A speed plateau or counter by itself does not identify the faulty component.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 1: Define What \u201cSlow\u201d Means<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">Record the affected task before running a benchmark. A slow webpage, an SFTP copy, and a backup job need different checks.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Write down:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The source and destination.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The protocol or application.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The direction of the transfer.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The observed rate and its unit.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">The time, date, and time zone.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Whether the issue affects one user, one destination, or every transfer.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Whether the problem is new and what changed before it started.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">For a website, separate slow response time from slow file delivery. A database query can delay the first byte even when the network has spare capacity.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Check Gbps, MB\/s, and MiB\/s<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Network plans usually show gigabits per second. File tools often show megabytes per second. There are eight bits in one byte.<\/span><\/p>\n<table>\n<thead>\n<tr>\n<th><b>Link rate<\/b><\/th>\n<th><b>Raw decimal byte rate before overhead<\/b><\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><span style=\"font-weight: 400;\">1 \u062c\u064a\u062c\u0627\u0628\u062a \u0641\u064a \u0627\u0644\u062b\u0627\u0646\u064a\u0629<\/span><\/td>\n<td><span style=\"font-weight: 400;\">125 MB\/s<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">10 \u062c\u064a\u062c\u0627\u0628\u062a \u0641\u064a \u0627\u0644\u062b\u0627\u0646\u064a\u0629<\/span><\/td>\n<td><span style=\"font-weight: 400;\">1,250 MB\/s<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><span style=\"font-weight: 400;\">These figures are conversions, not promised application results. Ethernet, IP, transport headers, and other overhead reduce the rate available for file data. MiB\/s uses a different unit from MB\/s, so compare like with like.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A download at 100 MB\/s is about 800 Mbps. That measurement alone cannot tell us whether storage, a client connection, or a network policy sets the limit.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Check What Your Plan Actually Provides<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Review the port speed, transfer allowance, traffic policies, and any committed bandwidth. \u201cUnmetered\u201d describes a billing arrangement. It does not remove endpoint limits or make every Internet path a 10Gbps path.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For a virtual machine, distinguish the host uplink from the guest&#8217;s allowed rate. A virtual adapter reporting a high speed does not prove that the VM has that capacity reserved.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">\u0644\u062f\u064a\u0646\u0627 <\/span><a href=\"https:\/\/atalnetworks.com\/ar\/10gbps-vs-1gbps-dedicated-servers\/\"><span style=\"font-weight: 400;\">10Gbps vs. 1Gbps dedicated server comparison<\/span><\/a><span style=\"font-weight: 400;\"> explains the capacity difference. Troubleshooting still requires measurements for the actual workload.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 2: Check Both Endpoints and the Active Link<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">A 10Gbps server cannot send a file at 10Gbps to a client whose connection tops out at 1Gbps. The same applies to an overloaded receiver or a remote service with a download cap.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Use a second endpoint with known spare capacity. A nearby test host helps reduce the number of network variables. Later, repeat the test against a remote endpoint that represents your users.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Identify the Interface Used by the Transfer<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Replace <\/span><span style=\"font-weight: 400;\">TEST_SERVER_IP<\/span><span style=\"font-weight: 400;\"> with the test endpoint&#8217;s address:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ip route get TEST_SERVER_IP<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ip -br link<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The route result identifies the selected interface for that destination. Policy routing, containers, tunnels, and multiple interfaces can make this different from the interface you expected.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For IPv6, use the matching IPv6 route lookup and keep the address family consistent across comparisons.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Next, replace <\/span><span style=\"font-weight: 400;\">INTERFACE<\/span><span style=\"font-weight: 400;\"> with the relevant interface name:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ethtool INTERFACE<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ethtool -i INTERFACE<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ip -s link show dev INTERFACE<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ethtool -S INTERFACE<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Check the reported speed, link state, driver information, and available error counters. Some commands need elevated permissions. Virtual interfaces may not expose the underlying physical link details.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A reported speed of <\/span><span style=\"font-weight: 400;\">10000Mb\/s<\/span><span style=\"font-weight: 400;\"> confirms the local link rate when the driver reports it correctly. It does not measure the complete path to a customer.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Compare Counters Before and After a Test<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Old errors can remain in counters long after an incident. Capture the values before the transfer, then compare them after it finishes.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Separate physical errors from other drops:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Rising CRC or frame errors can point toward a cable, optic, port, or signal problem.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Receive drops can involve queues, buffers, driver behavior, or a host that cannot process incoming traffic quickly enough.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Counter names and meanings vary by adapter and driver.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">For a dedicated server, ask the provider to check the switch side when the host counters suggest a physical issue. Include the test time so the provider can match its records.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">If the adapter or server hardware changed recently, also check the supported transceiver or cable type and the negotiated PCIe link width and speed. A capable NIC still needs enough bandwidth through its host connection.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 3: Run a Controlled Network Test<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">A normal iperf3 test helps separate network transfer performance from file reads and writes. It does not reproduce every part of an HTTPS download, VPN session, or backup job.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Use endpoints you own or have permission to test. Run high-rate tests in an agreed window because they can consume capacity used by customers. Start with a rate-limited test if spare capacity is uncertain.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The commands below are diagnostic examples, not published benchmark results.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Prepare the Test<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Record the installed version on both hosts:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">iperf3 &#8211;version<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Use compatible iperf3 installations. Starting with version 3.16, iperf3 uses a separate thread for each test stream. Older builds can behave differently under parallel load. See the <\/span><a href=\"https:\/\/software.es.net\/iperf\/faq.html\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">ESnet iperf3 FAQ<\/span><\/a><span style=\"font-weight: 400;\">.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">On the receiving test host, restrict TCP port 5201 to the source host in the firewall. Bind the listener to its intended local test address:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">iperf3 -s -B LOCAL_TEST_IP<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Replace the capitalized placeholders with real addresses. Do not expose an unrestricted test listener to the Internet.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Compare One Stream, Several Streams, and Both Directions<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Run these tests separately from the client. <\/span><span style=\"font-weight: 400;\">-R<\/span><span style=\"font-weight: 400;\"> makes the listening server send data back to the client. <\/span><span style=\"font-weight: 400;\">-P 4<\/span><span style=\"font-weight: 400;\"> uses four parallel streams. <\/span><span style=\"font-weight: 400;\">-J<\/span><span style=\"font-weight: 400;\"> saves JSON output. The examples use 30 seconds per test.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">iperf3 -c TEST_SERVER_IP -P <\/span><span style=\"font-weight: 400;\">1 <\/span><span style=\"font-weight: 400;\">-t <\/span><span style=\"font-weight: 400;\">30 <\/span><span style=\"font-weight: 400;\">-J &gt; forward-one.json<\/span><\/p>\n<p><span style=\"font-weight: 400;\">iperf3 -c TEST_SERVER_IP -P <\/span><span style=\"font-weight: 400;\">4 <\/span><span style=\"font-weight: 400;\">-t <\/span><span style=\"font-weight: 400;\">30 <\/span><span style=\"font-weight: 400;\">-J &gt; forward-four.json<\/span><\/p>\n<p><span style=\"font-weight: 400;\">iperf3 -c TEST_SERVER_IP -P <\/span><span style=\"font-weight: 400;\">1 <\/span><span style=\"font-weight: 400;\">-t <\/span><span style=\"font-weight: 400;\">30 <\/span><span style=\"font-weight: 400;\">-R -J &gt; reverse-one.json<\/span><\/p>\n<p><span style=\"font-weight: 400;\">iperf3 -c TEST_SERVER_IP -P <\/span><span style=\"font-weight: 400;\">4 <\/span><span style=\"font-weight: 400;\">-t <\/span><span style=\"font-weight: 400;\">30 <\/span><span style=\"font-weight: 400;\">-R -J &gt; reverse-four.json<\/span><\/p>\n<p><span style=\"font-weight: 400;\">These commands do not set a traffic cap. Use the <\/span><a href=\"https:\/\/software.es.net\/iperf\/invoking.html\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">installed iperf3 manual<\/span><\/a><span style=\"font-weight: 400;\"> to choose rate limits for your environment. Stop the listener with <\/span><span style=\"font-weight: 400;\">Ctrl+C<\/span><span style=\"font-weight: 400;\"> and remove the temporary firewall access when testing ends.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Record the receiving side&#8217;s throughput, retransmission information where available, and host metrics. Repeat an inconsistent result under the same conditions before treating it as a baseline.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Read the Results as a Pattern<\/span><\/h3>\n<table>\n<thead>\n<tr>\n<th><b>Test pattern<\/b><\/th>\n<th><b>What it suggests<\/b><\/th>\n<th><b>Next action<\/b><\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><span style=\"font-weight: 400;\">One and four streams both perform well<\/span><\/td>\n<td><span style=\"font-weight: 400;\">The tested endpoints and path can carry that load at that time<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Compare the real application&#8217;s storage and processing<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Four streams outperform one<\/span><\/td>\n<td><span style=\"font-weight: 400;\">The single-flow result has a limit that parallel traffic partly avoids<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Check TCP state, per-core CPU use, and flow distribution<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Both directions perform poorly<\/span><\/td>\n<td><span style=\"font-weight: 400;\">A shared endpoint or path limit remains possible<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Check host counters, CPU, configured caps, and a second endpoint<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Only one direction performs poorly<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Sending and receiving conditions differ<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Compare CPU, disk involvement, policies, and paths in each direction<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Nearby tests work; remote tests do not<\/span><\/td>\n<td><span style=\"font-weight: 400;\">The issue depends on the remote endpoint or wider path<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Compare RTT, TCP behavior, routing, and remote capacity<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><span style=\"font-weight: 400;\">A strong parallel result does not prove that one file transfer will reach the same rate. Parallel flows can use more processing resources and may follow different network paths.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Keep send and receive results separate. Adding them together does not describe one-way throughput.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 4: Check Storage and the Application<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">If iperf3 performs well but a file copy does not, inspect the parts that differ between the two tests.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">The sender must read the data. The receiver must process and store it. A slow stage can hold back the whole transfer.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Watch Disk Activity During the Actual Copy<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">On Linux, tools from the <\/span><span style=\"font-weight: 400;\">sysstat<\/span><span style=\"font-weight: 400;\"> package can show storage activity:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">iostat -xz <\/span><span style=\"font-weight: 400;\">1 5<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Review read and write rates, request latency, and queue behavior on both systems. Do not diagnose a modern SSD or RAID array from <\/span><span style=\"font-weight: 400;\">%util<\/span><span style=\"font-weight: 400;\"> alone. Device parallelism and workload type affect how that number relates to capacity.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Check for:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Backups, snapshots, or RAID rebuilds running at the same time.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">High write latency on the receiving system.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Many small files that require frequent metadata work.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Filesystem or device errors.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Storage quotas and low free space.<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">A fast first transfer followed by slower sustained transfers.<\/span><\/li>\n<\/ul>\n<p><span style=\"font-weight: 400;\">The last pattern can involve memory caching or a device&#8217;s write cache. A short burst does not establish the sustained rate.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Use a representative dataset and record whether it was cached. Avoid clearing production caches just to produce a benchmark. Do not run write tests against a raw device containing live data.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Check Protocol and Application Limits<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">SFTP, HTTPS, SMB, and backup software do different work. Encryption, compression, checksums, synchronous writes, and per-user limits can affect throughput.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For example, a large sequential file and a directory containing thousands of small files are different workloads. The directory copy may spend much of its time opening files and updating metadata.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">If compression is enabled, compare its CPU cost with its benefit for your data. Already compressed media often leaves little room for further size reduction. Keep required encryption and access controls in place while testing.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">A fast iperf3 result also does not rule out protocol-specific traffic policies. Your application may use a different port, proxy, tunnel, or destination from the test.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 5: Check CPU Use Per Core<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">Average CPU use can hide the bottleneck. On a many-core server, one busy core may limit a transfer while the system-wide percentage stays low.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">During the affected workload, run:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">mpstat -P ALL <\/span><span style=\"font-weight: 400;\">1 5<\/span><\/p>\n<p><span style=\"font-weight: 400;\">pidstat -u <\/span><span style=\"font-weight: 400;\">1 5<\/span><\/p>\n<p><span style=\"font-weight: 400;\">These tools also come from <\/span><span style=\"font-weight: 400;\">sysstat<\/span><span style=\"font-weight: 400;\">. Check both endpoints.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Look for a busy application worker, encryption process, or high soft-interrupt processing on individual cores. On a VM, CPU steal time can indicate that the guest is waiting for host CPU time.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Check Packet Processing and Queue Distribution<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">A NIC can distribute incoming flows across receive queues. Linux Receive Side Scaling and related mechanisms help spread work across CPUs. The result depends on the adapter, driver, traffic, and configuration. The <\/span><a href=\"https:\/\/docs.kernel.org\/networking\/scaling.html\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">Linux kernel networking guide<\/span><\/a><span style=\"font-weight: 400;\"> explains these relationships.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Inspect supported queue information and offload settings:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ethtool -l INTERFACE<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ethtool -k INTERFACE<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Some adapters do not support these queries. A missing feature or command error does not by itself prove a fault.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Do not turn every offload off or set every queue count to its maximum. First identify which CPU or queue is overloaded. Then choose a change that fits the hardware and measure its effect.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Account for Packet Size and Middleboxes<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">The same bit rate requires more packets per second when packets are smaller. A server that handles large bulk transfers well may struggle with a packet-heavy VPN, game, or proxy workload.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Firewalls, NAT, virtual switches, tunnels, and traffic inspection add work between the application and physical NIC. Test the path your users actually take. A successful test outside a VPN does not measure encrypted tunnel capacity.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 6: Examine TCP Behavior, Loss, and Latency<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">TCP throughput depends partly on how much data can remain unacknowledged while the sender waits for replies. Higher round-trip time makes this more demanding. Loss and limited windows can also reduce throughput. ESnet describes these links in its <\/span><a href=\"https:\/\/fasterdata.es.net\/network-tuning\/tcp-issues-explained\/\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">TCP performance guide<\/span><\/a><span style=\"font-weight: 400;\">.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Inspect active TCP connections during the transfer:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ss -ti<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Match the output to the test connection. Depending on the kernel and connection state, it can show RTT, congestion-window information, and retransmission details.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Use the Bandwidth-Delay Product as a Check<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">The bandwidth-delay product estimates the amount of data needed in flight at a target rate:<\/span><\/p>\n<p><b>Bytes in flight = target bits per second \u00d7 round-trip time in seconds \u00f7 8<\/b><\/p>\n<p><span style=\"font-weight: 400;\">For a 10Gbps target:<\/span><\/p>\n<table>\n<thead>\n<tr>\n<th><b>RTT<\/b><\/th>\n<th><b>Calculated data in flight<\/b><\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><span style=\"font-weight: 400;\">10 ms<\/span><\/td>\n<td><span style=\"font-weight: 400;\">12.5 MB<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">40 ms<\/span><\/td>\n<td><span style=\"font-weight: 400;\">50 MB<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">100 ms<\/span><\/td>\n<td><span style=\"font-weight: 400;\">125 MB<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><span style=\"font-weight: 400;\">These are decimal calculations before protocol overhead. They are not instructions to assign those values to every socket buffer. Application behavior, receive-window limits, congestion control, and operating-system memory management also matter.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Interpret Retransmissions Carefully<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Rising retransmissions during a slow transfer deserve investigation. They can reflect lost data, but reordering and spurious retransmissions can complicate the picture.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Possible loss sources include congestion, an overloaded receiver, a firewall, and physical faults. ESnet&#8217;s <\/span><a href=\"https:\/\/fasterdata.es.net\/performance-testing\/troubleshooting\/packet-loss\/\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">packet-loss troubleshooting reference<\/span><\/a><span style=\"font-weight: 400;\"> covers these causes.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Compare retransmissions with interface counters, CPU use, and the transfer timeline. A retransmission total cannot identify the router or cable responsible.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Also compare latency at idle and under load. If latency climbs while a transfer fills the connection, queues may be building. More bandwidth or larger buffers are not automatic fixes for every queueing problem.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 7: Check MTU and Routing<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">The maximum transmission unit, or MTU, sets the largest IP packet an interface can send without the relevant fragmentation behavior. Tunnels add headers and can reduce the usable size for traffic inside them.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Investigate MTU When the Symptoms Fit<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">MTU trouble is worth checking when small requests work but larger transfers stall, especially after a VPN, tunnel, or network change.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Check the local interface MTU:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">ip link show dev INTERFACE<\/span><\/p>\n<p><span style=\"font-weight: 400;\">If available, use <\/span><span style=\"font-weight: 400;\">tracepath<\/span><span style=\"font-weight: 400;\"> to gather path information:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">tracepath TEST_SERVER_IP<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Filtering can prevent useful replies. A failed probe is not enough to prove an MTU mismatch.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Do not set MTU 9000 across a public Internet path as a general speed fix. Jumbo frames need support across the relevant network segment. For a tunnel, check its overhead, the inner MTU, and whether required path-MTU signals can reach the sender.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">If you change an interface that carries your management connection, keep console access and a clear recovery method available.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Compare Routes Without Blaming the First Silent Hop<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">A bounded MTR report can help document a route:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">mtr -n -r -c <\/span><span style=\"font-weight: 400;\">50 <\/span><span style=\"font-weight: 400;\">TEST_SERVER_IP<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Record the test time and protocol. Default diagnostic probes may receive different treatment from application traffic.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">An intermediate router can limit replies to probes while forwarding traffic normally. Loss that appears at one hop and disappears at later hops does not establish forwarding loss at that router. Cloudflare&#8217;s <\/span><a href=\"https:\/\/www.cloudflare.com\/learning\/network-layer\/what-is-mtr\/\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">MTR explanation<\/span><\/a><span style=\"font-weight: 400;\"> shows this distinction.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Even missing destination replies need context. Compare them with application failures, retransmissions, and endpoint behavior. Collect reports from both ends when possible because the return path may differ.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">For a useful routing comparison, keep the workload and settings fixed. Change only the destination or test time. A single public speed-test result is weak evidence because its endpoint load and selected route can change.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">Step 8: Apply One Fix and Repeat the Original Test<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">A useful correction addresses the measured limit. It should improve the affected workload without creating another problem.<\/span><\/p>\n<table>\n<thead>\n<tr>\n<th><b>Evidence<\/b><\/th>\n<th><b>Possible corrective action<\/b><\/th>\n<th><b>What to measure afterward<\/b><\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><span style=\"font-weight: 400;\">A local link reports 1Gbps where 10Gbps is expected<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Correct the port configuration or incompatible physical component<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Link rate, error counters, and throughput<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Physical error counters rise during transfers<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Have the provider inspect the cable, optic, and switch port<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Error-counter changes during the same load<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Disk latency rises as file throughput falls<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Address storage contention or the storage design<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Sustained application transfer rate and disk latency<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">One application worker reaches its CPU limit<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Review supported worker settings or CPU requirements<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Per-core use and application response time<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Host drops rise with packet-processing load<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Review queues, CPU distribution, and supported driver settings<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Drops, CPU use, and packet-heavy workload behavior<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">Only one remote path performs poorly<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Share comparable tests with the relevant providers<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Results to the affected destination at matching times<\/span><\/td>\n<\/tr>\n<tr>\n<td><span style=\"font-weight: 400;\">A documented traffic cap matches the observed ceiling<\/span><\/td>\n<td><span style=\"font-weight: 400;\">Adjust the policy or service capacity if required<\/span><\/td>\n<td><span style=\"font-weight: 400;\">The same test within the revised allowance<\/span><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p><span style=\"font-weight: 400;\">Before each change, save the current configuration and record how to restore it. Change one variable at a time. If the result worsens or fails to improve, restore the previous state before trying another change.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Repeat the original application task as well as the network test. Better benchmark numbers are not enough if users still face slow downloads or timeouts.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">What to Send Your Hosting Provider<\/span><\/h2>\n<p><span style=\"font-weight: 400;\">A clear support request helps the team reproduce the problem and compare its network records with yours.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Use this template:<\/span><\/p>\n<p>&nbsp;<\/p>\n<p><span style=\"font-weight: 400;\">Server identifier:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Affected application or protocol:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">When the issue started:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Test date, time, and time zone:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Source and destination addresses:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Traffic direction:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Expected result and basis for that expectation:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Observed result and units:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Test commands and tool versions:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Single-stream and parallel-stream results:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Nearby and remote endpoint results:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Per-core CPU observations:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Storage observations:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Interface-counter changes during the test:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">TCP or route observations:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Recent changes:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Business impact:<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Send diagnostic files through the provider&#8217;s private support channel. Remove credentials, tokens, and unrelated customer data. Packet captures can contain sensitive information, so agree on their scope before collecting or sharing them.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">If you host with Atal Networks, include these details in your support request. If measurements show that your workload needs a different hardware or network allocation, review our <\/span><a href=\"https:\/\/atalnetworks.com\/ar\/10gbps-server\/\"><span style=\"font-weight: 400;\">10Gbps dedicated server options<\/span><\/a><span style=\"font-weight: 400;\"> and discuss the measured requirement with our team.<\/span><\/p>\n<h2><span style=\"font-weight: 400;\">\u0627\u0644\u0623\u0633\u0626\u0644\u0629 \u0627\u0644\u0634\u0627\u0626\u0639\u0629<\/span><\/h2>\n<h3><span style=\"font-weight: 400;\">Why am I getting about 1Gbps on a 10Gbps server?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">A client connection, intermediate link, traffic policy, or endpoint resource may limit the transfer. Check both link rates and compare a controlled network test with the real application. A result near 1Gbps does not identify the limiting component by itself.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Should a 10Gbps server transfer files at 1,250 MB\/s?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">That figure is the raw bit-to-byte conversion before overhead. Real file throughput depends on the protocol, path, storage, CPU, and receiving system. Set an acceptance target for your specific workload and test conditions.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Why is iperf3 fast while SFTP is slow?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">SFTP includes file reads, writes, encryption, and application processing that a normal iperf3 test does not reproduce. Check disk latency and per-core CPU use during the copy. Also check whether both tests use the same route and traffic policy.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Why do parallel connections improve throughput?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Several flows can use more processing resources and avoid a limit affecting one flow. They may also take different paths through load-balanced equipment. The result shows aggregate performance for those flows, not the speed available to every single connection.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Does distance make 10Gbps impossible?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Distance increases latency, which raises the amount of data needed in flight to sustain a target rate. Loss, endpoint capacity, and TCP behavior also matter. Distance alone is not a fixed throughput ceiling.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Will jumbo frames fix a slow 10Gbps connection?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Only consider them where the network supports them and measurements justify the change. An inconsistent MTU can cause failures. For Internet traffic, investigate path MTU and tunnel overhead before changing interface settings.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Does MTR packet loss prove that my provider is dropping traffic?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">No. A router may limit diagnostic replies while forwarding application traffic normally. Compare later hops, endpoint behavior, retransmissions, and tests from both directions before assigning a cause.<\/span><\/p>\n<h3><span style=\"font-weight: 400;\">Should I upgrade the server before troubleshooting?<\/span><\/h3>\n<p><span style=\"font-weight: 400;\">Measure the bottleneck first. More CPU can help a CPU-bound workload, and faster storage can help a storage-bound workload. Neither change fixes a slow client connection or an unrelated remote path limit.<\/span><\/p>","protected":false},"excerpt":{"rendered":"<p>A 10Gbps server can be slow because the client, storage, CPU, application, or network path cannot sustain the requested rate. [&hellip;]<\/p>\n","protected":false},"author":2,"featured_media":25246,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"site-sidebar-layout":"default","site-content-layout":"","ast-site-content-layout":"default","site-content-style":"default","site-sidebar-style":"default","ast-global-header-display":"","ast-banner-title-visibility":"","ast-main-header-display":"","ast-hfb-above-header-display":"","ast-hfb-below-header-display":"","ast-hfb-mobile-header-display":"","site-post-title":"","ast-breadcrumbs-content":"","ast-featured-img":"","footer-sml-layout":"","ast-disable-related-posts":"","theme-transparent-header-meta":"default","adv-header-id-meta":"","stick-header-meta":"","header-above-stick-meta":"","header-main-stick-meta":"","header-below-stick-meta":"","astra-migrate-meta-layouts":"set","ast-page-background-enabled":"default","ast-page-background-meta":{"desktop":{"background-color":"var(--ast-global-color-4)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"ast-content-background-meta":{"desktop":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"tablet":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""},"mobile":{"background-color":"var(--ast-global-color-5)","background-image":"","background-repeat":"repeat","background-position":"center center","background-size":"auto","background-attachment":"scroll","background-type":"","background-media":"","overlay-type":"","overlay-color":"","overlay-opacity":"","overlay-gradient":""}},"footnotes":""},"categories":[1],"tags":[],"class_list":["post-25245","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-enterprise-grade-server"],"acf":[],"_links":{"self":[{"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/posts\/25245","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/comments?post=25245"}],"version-history":[{"count":1,"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/posts\/25245\/revisions"}],"predecessor-version":[{"id":25247,"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/posts\/25245\/revisions\/25247"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/media\/25246"}],"wp:attachment":[{"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/media?parent=25245"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/categories?post=25245"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/atalnetworks.com\/ar\/wp-json\/wp\/v2\/tags?post=25245"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}