Skip to content
Documentation menu
Products & services
Back to section

VPS DDoS protection and traffic statistics

Choose Auto, On or Off, confirm when protection is active, configure country and per-port TCP session limits, and interpret traffic counters, SYN challenges and sampled observations.

Services / VPS security

Choose protection settings and understand their effect

Learn when to use Auto, On or Off, confirm activation, configure country and TCP session limits, and read traffic observations accurately.

Confirm the applied state

A saved setting can still be preparing or waiting for the host. Check the individual status before treating a control as active; monitoring samples are not complete connection logs.

What protection does and when to use it

A denial-of-service attack tries to make a service unavailable by exhausting network, connection or application resources. The VPS protection controls help you manage supported network flood mitigation, reduce unwanted access by country, limit TCP connection pressure and inspect traffic for your allocated addresses.

Use these controls alongside firewall access rules, software updates and application monitoring. A protection mode cannot guarantee uninterrupted service against every attack, recover an overloaded application or correct a guest networking error. Website-specific HTTP request filtering belongs to Website Protection/WAF.

Open your VPS, choose Firewall, and use DDoS protection for settings or Statistics for traffic. Features depend on the host, address family and current setup. A disabled control or a pending status is not evidence that its protection is already enforcing.

Choose a DDoS mode

Mode What it does When to use it
Auto Monitors traffic, activates supported mitigation during a detected flood and returns to monitoring after recovery. A starting point for an eligible public service that needs automatic response without forcing mitigation continuously.
On Requests the supported TCP/SYN protection continuously, once setup is confirmed. When you intentionally want that protection active during normal traffic as well as an incident.
Off Turns off the selectable automatic/forced DDoS mode while traffic monitoring continues. When you deliberately do not want that mode enabled. Other independently configured controls still need separate review.

Where supported, TCP SYN protection challenges connection attempts before allowing them to establish a connection to the server. It helps reduce incomplete TCP-handshake pressure. It is not a browser CAPTCHA and does not analyse the content of an HTTPS request.

TCP/SYN setup and the per-port session controls described below are qualified for supported IPv4 paths. Do not assume that selecting a mode establishes equivalent protection for every IPv6 address, UDP service or application-layer workload. Check the availability and applied state shown for your VPS.

Set the mode and confirm activation

  1. Select DDoS protection and review the current Protection mode badge.
  2. Choose Auto, On or Off from DDoS protection mode, if available.
  3. Select Save in that setting's section. The sections have independent save controls; saving the mode does not submit unsaved country or session-limit edits.
  4. Wait for the applied state and any setup work. Refresh the page and test a normal connection to a listening TCP service.
  5. Check the badge and Statistics during the relevant traffic interval. Do not generate an attack to test the feature.
Status Meaning and next step
Not saved The selection is still a local edit. Save it or discard it.
Propagating the rule / Synchronization pending The requested setting has not yet been confirmed on the host.
Preparing protection / Waiting for TCP traffic Setup is not complete. Use ordinary traffic to your real service and check again.
Waiting for VPS to start The VM must be running for setup to proceed; start it only when appropriate.
Waiting for host / TCP setup needs review Setup cannot currently be confirmed. Record the state and contact support if it persists.
Monitoring Auto is monitoring; it does not mean mitigation is currently active.
Mitigating Auto reports active mitigation for the current condition.
Forced On is applied and the required setup is confirmed.
Off The selectable DDoS mode is off; monitoring can continue.

Saved is not the same as active. Read the setup status as well as the selected mode. Do not repeatedly reboot or reinstall a VPS to force protection readiness.

Settings remain independent

Country restrictions, TCP session limits and edge/VM access rules are configured separately. Setting DDoS protection to Off does not automatically remove them. TCP session admission can remain active independently of Auto's current mitigation state.

The workspace's overall Enforcement enabled badge can also reflect other active controls, including supported malicious-network filtering. For a particular feature, read its own applied-state indicator rather than treating the overall badge as proof of its state.

Baseline network checks

Where supported and active, platform checks can also reject invalid TCP traffic and filter addresses classified by the maintained malicious-network feed. These help reduce unwanted traffic before it reaches the guest. They are distinct from your own port allowlist and from Auto's current flood state; not every platform safeguard has a customer-editable threshold.

Use the displayed block reasons and availability notices to understand what acted. If legitimate traffic appears affected, provide support with the source/destination addresses, port, time and classification instead of disabling unrelated controls.

Country restrictions

Why: Reduce traffic from countries that should not use a particular service, or limit a private-purpose service to a known geographic audience.

  1. Under Country restrictions, choose Allow all countries, Block selected countries, or Allow selected countries only.
  2. For a selected-country policy, choose the countries you intend. The selector supports up to 16 countries.
  3. Review the locations of administrators, legitimate visitors, monitoring systems, VPN exits and upstream proxies/CDNs before saving.
  4. Select Save in the country section, wait for the applied state and test from an allowed connection.

Allow selected countries only blocks other countries and unknown locations. Include the country used by your administration connection. Country matching uses IP-allocation data, not a person's physical location or identity; VPNs, mobile networks and proxies can change the location seen by the filter.

A geographic restriction can affect the proxy or CDN address that connects to your server rather than the final visitor. Do not assume that an Allow rule in a different firewall layer bypasses a country restriction. If the host does not offer country filtering, leave the unavailable control alone and ask support about eligibility.

TCP sessions per source IP

Why: Limit how many simultaneous TCP connections a single source can hold to a selected service. This can help with connection exhaustion or abusive clients while allowing each service a different limit.

  1. Under TCP session limits, choose Add port.
  2. Enter the service's destination port and its Maximum concurrent TCP sessions per source IP. You can configure up to three different ports.
  3. Use a whole-number limit from 0 to 10,000. 0 disables that port's session limit; it does not block every connection to the port.
  4. Select Save in the session section. Wait for Active when a nonzero limit is intended, or Disabled when limits are off. Not ready, Waiting for host and Propagating the rule need further confirmation.
  5. Test legitimate new connections and monitor session rejections. Adjust using your actual workload rather than an arbitrary low number.

Limits apply per source IP and destination IPv4/port, not per account or person. Office NAT, shared Wi-Fi, a reverse proxy or a CDN can place many legitimate clients behind one source address. HTTP/2 and connection pools can also handle many requests through a small number of TCP connections. A TCP-session limit is not an HTTP request-rate limit or a UDP limit.

If the page shows an existing all-port session limit, it remains in effect until you save a port-specific configuration. Review the intended replacement before saving; a three-port configuration should not be assumed to preserve an older all-port scope.

Understand traffic statistics

Why: Establish whether traffic is reaching a protection stage, which controls are acting, and whether the timing matches an application problem.

  1. Open Statistics. Select one allocated IP or all IP addresses, then choose inbound, outbound or both directions.
  2. Read the latest sample timestamp and any availability or coverage notice.
  3. Give a live-rate chart time to collect two valid samples. The live chart builds during the page session; an empty first chart is not proof of zero traffic.
  4. Compare counters, protocol/port observations and protection activity with a real application test at a recorded time.
Measurement How to interpret it
Received Packets or bytes observed at that protection stage.
Passed Traffic allowed onward by that stage. Later firewall stages or the guest can still reject it.
Blocked Traffic actually dropped at the reported stage.
Audit matches Matches counted without enforcing that edge-policy block. Do not count these as blocked packets.
SYN challenges and validations Packet events in the TCP protection process, not unique attackers, users or completed application requests.
TCP session rejections Later-stage admission decisions. They are reported separately and must not be added blindly to edge blocked totals.

Packet counters and bytes are cumulative within the displayed collection/allocation scope; PPS means packets per second and needs a measured interval. Packets per second and bandwidth measure different pressure: many small packets can consume processing capacity without using much bandwidth.

Inbound local ports are destination ports; outbound local ports are source ports. TCP flags can overlap on the same packet. Counter resets, collection gaps, unavailable values and stale timestamps require care when comparing two readings. Same-host traffic can be outside edge observation, and outbound packets can be counted before segmentation, so these counters are not a complete wire-level packet capture.

Source observations and block reasons

Source observations are sampled records from the displayed recent window. Filters can show blocked, forwarded, SYN-challenge or Audit samples. Sample counts are not packet totals and an empty table does not establish that nobody connected or that no packets were blocked.

Rate-limited samples, buffer losses and reader errors describe collection completeness. Retaining samples for later display does not turn a sampled stream into a complete connection or request log. Port tracking and rankings also have capacity and coverage limits; inspect the notices on the page.

Edge block reasons explain the primary classification attached to an observed drop, such as an access rule, country restriction, malicious-network classification or TCP validation. A TCP-validation label alone does not prove a particular attack type. A country or reputation match does not identify a person or prove a successful compromise.

Protection activity shows observed mode changes and mitigation periods. Compare it with timestamps and service tests; it is not, by itself, a confirmed attack history. Provider safeguards can operate independently of customer-configured settings; use the available controls rather than assuming every internal threshold is editable.

DPI and website attacks

The DPI tab currently describes unavailable inspection features. It is not an active application-protocol firewall and does not turn on inspection by opening the tab. Use only controls whose applied status the portal confirms.

For HTTP exploit filtering, website access rules and proxy/SSL controls, use DNS and website protection. Continue maintaining the guest OS and application even when network protection is enabled.

If the service is slow or unreachable

  1. Confirm the VPS is running and the application is listening. Check the guest's CPU, memory, storage and application logs.
  2. Review the applied DDoS mode, setup status, country policy, session limits and firewall rules.
  3. Check whether the real administration/client source IP is permitted, including any VPN, NAT or proxy address.
  4. Compare fresh counters and the application test. A quiet graph, a Passed counter or a missing sample is not a complete diagnosis.
  5. Reverse only an identified problematic setting through its own Save control. Record errors and contact support when protection or synchronization remains unavailable.

Send support the service ID, affected IPv4/IPv6 address, source IP, protocol/port, exact time and timezone, current applied states and a description of the failed test. Do not include private keys, passwords or unnecessary request/payment data.

Return to VPS firewall rules and policies or Manage a VPS.