Multi-homed uplinks
Several transit providers per edge reduce single-carrier failure domains. During maintenance, traffic may shift AS paths—your monitoring should tolerate brief reconvergence.
Connectivity
Network & delivery you can plan around
Great hosting is not a single postcode—it is routing, caching, SSL, and how your stack behaves under real traffic. This page explains how HostStack approaches regions, edge security, and measurement so finance and engineering read the same brief.
Growing network of live sites, stores & WordPress installs on NVMe-backed plans.
- 30-day money-back on eligible plans
- INR checkout & GST-ready invoices
- UPI, cards & net banking
Executive brief
Network at a glance
This document is written for engineering and procurement teams who need routing context before they sign. Numbers are fleet-level targets unless your order defines something more specific; contractual SLAs remain in terms and product addenda.
What we optimise for
- Predictability — stable paths, monitored DNS, and disciplined change windows rather than best-effort improvisation.
- Observability — curated status messaging, ticket-based escalation, and optional premium monitoring on qualified lines.
- Honest bounds — we do not pretend a single benchmark IP represents every footprint; see the measurement section for a fair recipe.
Operational targets
- Service availability goal
- 99.9% monthly
- Core capacity class
- Multi-gig uplinks
- Support ethos
- 24/7 human queue + escalation
- IPv6 stance
- Enabled where SKU and upstream allow—confirm during purchase or onboarding.
| Layer | Hosted responsibility | Your responsibility |
|---|---|---|
| Physical & power | Facility power and cooling tiers purchased from upstream DC operators; redundancy claims follow their design. | Platform choice, failover architecture, snapshots, backups, and restores. |
| WAN / transit | Multi-homed carrier diversity, upstream DDoS scrubbing tiers, BGP maintenance within policy windows. | CDN selection, registrar DNS, TTL strategy, firewall rules on VPS, rate limits for APIs. |
| Application | Managed-stack tuning on eligible lines (PHP opcode, caches, LiteSpeed profiles where bundled). | Code paths, queues, cron storms, inefficient SQL, unsecured forms. |
Request lifecycle
End-to-end delivery path
Visitors rarely experience “raw DC ping” alone. The following stack is simplified but matches how procurement teams evaluate us against CDNs and origin hosts.
-
Recursive DNS & resolver path
Client stub resolver → authoritative DNS chain (often your registrar plus optional secondary DNS). Misconfigured TTLs or stale glue records break experiences before packets ever reach HostStack.
-
Edge / CDN (optional but common)
Caches, TLS termination, compression, bot filtering. Even with LiteSpeed on-origin, pairing a CDN front for static bursts often yields better UX than brute-forcing a single egress.
-
Internet transit → facility edge
Multi-homed carriers, scrubbing policies, BGP advertisements, and sane burst capacity. Failures here look like asymmetric routing or packet loss—not slow PHP.
-
Application stack
Opcode caches, queues, cron load, WooCommerce sessions, VPS firewall rules—you still own optimisation here even when we tune the substrate.
Peering & capacity
Backbone, transit & routing discipline
We contract with facilities and carriers that maintain redundant power, diverse fiber entrances, and scrubbing fabric sized for shared-hosting economics. That does not mean infinite clean bandwidth—plan upgrades when your observability shows sustained congestion.
BGP hygiene
Prefix advertisements follow registry data. If you bring your own IP space to custom builds, we validate IRR/RPKI records before production.
Operational windows
Planned reroutes roll during published windows whenever possible—follow status notices and subscriber emails for advanced warning.
| Symptom | Likely plane | First response |
|---|---|---|
| Elevated TTFB on HTML only | PHP / datastore / cron | Profile queries, opcode cache hits, WooCommerce cron; open ticket with timelines + URLs. |
| Regional packet loss traceroute stops mid-way | Transit / ISP | Collect MTR from two ISPs & share with support; escalate if coordinated maintenance not posted. |
| Complete drop on CDN edge | Edge DNS or WAF | Bypass CDN briefly to isolate origin vs edge; purge stale rules causing challenge loops. |
Operational defaults teams expect from HostStack
Reliability
99.9% uptime target
Resilient DNS, monitored rollbacks, and clear escalation paths—track live checks on status.
Edge security
Up to 17 TB-tier mitigation
Network-layer mitigation is provisioned behind our edge, with scrubbing tiers staged up to 17 TB aggregate attack-volume absorption on contracted upstream capacity. Eligible scope follows your SKU, point of presence, carrier classification, and published acceptable-use terms; pair high-risk workloads with CDN or WAF where appropriate. Layer application traffic with CDN or WAF where it makes sense.
Finance-ready
INR-first billing
UPI, cards, and net banking through established processors—download receipts from the client area for GST workflows.
Operations
24/7 human support
Real engineers on tickets and chat when routing, SSL renewals, or migrations need attention—not script-only bots.
Regions & workloads
Built for India. Ready when you need global VPS.
HostStack is operated from Kolkata, West Bengal. Shared, cloud-style, and WordPress plans prioritise fast delivery to Indian visitors with NVMe-backed storage and CDN-ready stacks. For KVM VPS, you select region at order time in the client area—align footprint with latency, data handling, and how you snapshot and restore.
Web & WordPress
SSL lifecycle, PHP tuning, and headroom that scales with traffic—upgrade paths stay explicit in the product line.
Commerce & WooCommerce
Treat checkout and admin URLs as first-class routes: caching, images, and edge behaviour matter as much as raw CPU.
VPS, workers & APIs
Predictable compute and networking for databases and automation—review VPS plans and pair with observability after cutover.
| Workload | Primary concern | Instrumentation |
|---|---|---|
| Corporate brochure HTML | CDN cache TTL + image weight | Synthetic probes from Indian metros + mobile |
| Headless storefront | API latency spikes & cart sessions | APM + origin vs edge waterfall comparison |
| Trading / SaaS VPS | Jitter towards venue/exchange hops | ICMP + QUIC/TCP pings from client vantage points |
| Realtime games | UDP loss & asymmetric routing | Server tick graphs + isp-specific capture |
Latency checks & test targets
We do not list static “benchmark” IP addresses here—they churn with inventory, vary by SKU, and create false confidence. For a serious pre-sales check, email contact with region, product interest, and sample client networks (e.g. broadband + 4G/5G). We will suggest appropriate traceroute or ping targets where policy allows.
Self-serve documentation: Knowledge base · Support · POP guide
| Signal | Suggested tool class | Interpretation caveat |
|---|---|---|
| ICMP RTT | Ping / smokeping | Deprioritised vs TCP on many routers; directional links may diverge. |
| HTTP(S) timing | curl -w, Lighthouse, Sitespeed | Watch DNS + TLS handshake separately from TTFB; warm caches distort. |
| Path trace | mtr | Some hops purposely deprioritise TTL expires—focus on sustained loss downstream. |
Security posture
DDoS layering, UDP workloads & fair use
Mitigation stacks differ by threat type. Volumetric floods are handled closest to inbound carriers; application-layer attacks require rate rules, caches, WAF policies, or challenge pages; UDP heavy traffic (gaming, realtime voice) exposes different signatures than HTTPS retail sites.
Layers 3–4
High packet-per-second spikes and spoofed sources are diverted into scrubbing fabric where possible—expect brief reroutes during major events.
Layer 7 HTTP(S)
Bots hitting expensive endpoints deserve edge challenge or cache rewrite before they starve PHP workers—even “unlimited bandwidth” slogans rarely save you from abusive query patterns.
UDP & realtime
Game panels and VPS lines can request tuned templates; still provision headroom because reflective floods may force temporary null-routes if abuse overwhelms fairness budgets.
Upstream carriers provide disciplined rate limits—but no operator can neutralise every attack geometry at unlimited scale. Pair HostStack origins with CDN offload, authenticated APIs, webhook signing, observability paging, and pre-written stakeholder comms.
- 1 Measure from real users — a single synthetic probe from one metro rarely matches mobile or last-mile behaviour.
- 2 Treat DNS as infrastructure — TTL, failover, and registrar locks affect incidents as much as raw ping.
- 3 Layer defences intentionally — origin hardening plus edge WAF/CDN still matters for storefronts and APIs.
Common questions
How should I measure latency before buying?
Test from the networks your audience actually uses—home Wi‑Fi and mobile data—not only your office line. Combine traceroute or ping with TTFB on real URLs once DNS and certificates are stable. If you need official targets, contact sales with your region and workload.
Why are public test IPs not listed here?
Addresses rotate between deployments and product lines. Publishing a static IP misleads buyers and support. Use the status page for operational visibility and open a ticket when you need a vetted target for procurement.
How do VPS regions work at HostStack?
KVM VPS plans let you choose region during checkout in the client area. Pick the footprint that satisfies latency, compliance, and backup strategy—then validate with monitoring after go-live.
What about DDoS or abuse traffic?
We use upstream filtering and fair-use policies aligned with industry practice. High-risk endpoints should still add CDN or WAF layers and maintain incident runbooks—no shared host replaces architectural discipline.
Can we get BGP visibility or sessions?
Commercial shared stacks do not include customer BGP sessions. Custom enterprise builds may negotiate advanced networking—bring ASN, prefixes, IRR/RPKI validation, and required RPO/RTO assumptions to pre-sales.
Is IPv6 available everywhere?
IPv6 rollout depends on product SKU and upstream handoff policy. Confirm on the provisioning ticket or invoice—if parity between v4/v6 endpoints is mandatory for compliance, cite that during procurement.
How should UDP-heavy workloads plan redundancy?
Run synthetic jitter tests from Tier-2/3 ISP eyeballs across India; keep failover messaging for players/customers scripted; optionally split regions if one transit path dominates your audience geography.
Deploy with clarity
Choose a plan, validate with data after cutover, and scale when your graphs—not a brochure—say it is time.