> For the complete documentation index, see [llms.txt](https://docs.console.zenlayer.com/welcome/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.console.zenlayer.com/welcome/elastic-compute/networking/01-overview/02-network-types.md). # Network Types Every EIP picks a **network type** at creation time. Network type is fixed for the lifetime of that object — switching network types means creating a new EIP on the other network and cutting over. It matters for billing because **different networks have different per-Mbps prices**. It matters for performance because the network type decides which autonomous system actually carries your packets between the POP and the rest of the internet. The available values are: | Network Type | Who operates it | Typical use | | ---------------- | ----------------------------------------------------- | ------------------------------------------------ | | **Premium BGP** | Zenlayer AS21859 + AS4229 backbone | Real-time, interactive, low-latency user traffic | | **Standard BGP** | Zenlayer, blended transit (Cogent, regional carriers) | CDN origin egress, VPN, backup, bulk traffic | The console shows only the network types available in your selected region. *** ## Zenlayer BGP — Standard vs. Premium Zenlayer operates the `BGP` network type in two tiers. Both are ZEC-native; the difference is which autonomous systems carry your packets. ### Premium BGP — AS21859 edge + AS4229 backbone The premium tier uses **AS21859** as the public-facing transit ASN at the POP, and **AS4229** as the private inter-POP backbone. The split is visible in public routing data, and it explains where the latency and reach advantages come from: | Source | AS21859 (peering edge) | AS4229 (backbone) | | --------------------------------------- | ----------------------------------------------------------------------------------------- | ----------------------------------------------------------------------- | | PeeringDB declared traffic | 10–20 Tbps | 20–50 Tbps | | PeeringDB declared IPv4 / IPv6 capacity | 20 000 / 10 000 prefixes | 100 000 / 50 000 prefixes | | Public peering exchanges | \~170 | \~24 (selective, with hyperscalers) | | Scope | Global | Global | | Traffic ratio | Heavy outbound | Heavy outbound | | Peering policy | Selective | Selective | | Network type (PeeringDB) | Content | Content | | Upstreams observed | Arelion (1299), Cogent (174), NTT (2914), Tata (6453), Lumen (3356) | Arelion (1299), PCCW Global (3491), NTT (2914), GTT (3257), Zayo (6461) | | Notable direct peers | 800+ peers — regional ISPs across APAC/EMEA (Bharti Airtel, Viettel, Etisalat, many more) | Google, AWS, Microsoft, Cloudflare, Meta, Apple, Netflix, ByteDance | Read that table as two layers of one product: * **AS4229 is where the hyperscaler peering lives.** It is present at fewer public IXes (\~24) but those are the high-value ones where Google/AWS/Microsoft/Meta/Apple/Netflix/ByteDance peer directly. Traffic to those destinations lands in one AS hop. * **AS21859 is the broader peering fabric.** It joins \~170 exchanges and peers with hundreds of regional ISPs — the networks that actually carry last-mile traffic to users in Southeast Asia, the Middle East, Africa, and LATAM. What that buys you on the wire: * Lower median RTT for most destinations reachable via a direct peer. * Tighter p95/p99 latency, since backbone transport does not share fate with the public internet between POPs. * Shorter AS paths to APAC and EMEA consumer networks. **Pick Premium BGP for:** real-time and interactive workloads — gaming, VoIP, trading, SaaS APIs, user-facing web traffic where tail latency matters. ### Standard BGP — Zenlayer-operated, blended upstreams Standard BGP is still Zenlayer's network — same POP, same EIP primitive, same set of billing methods — but the path under the EIP is built from **more cost-effective upstreams and transit arrangements**. You give up some of the peering density and path length that Premium gets from AS4229 + AS21859; in exchange the per-Mbps price is materially lower. Characteristics: * Uses a blend of commercial transit in addition to private peering. * Routing typically takes more AS hops for some destinations compared with Premium. * Connectivity and latency are not as tight — the degradation is most visible on consumer last-mile networks in Asia-Pacific. * Per-Mbps price is noticeably lower than Premium. **Pick Standard BGP for:** CDN origin egress, VPN backhaul, backup and replication, batch ETL, software distribution, log/telemetry egress — any workload where bytes dominate and an extra few milliseconds of RTT does not change the user experience. ### Regions where Standard BGP is available Standard BGP is currently exposed on these ZEC regions: | Area | Region | POP | | ---------------------- | ------------------ | ---------------- | | Americas | `na-west-1` | Los Angeles | | Americas | `na-central-2` | Dallas | | Americas | `na-east-1` | Washington, D.C. | | Americas | `na-south-1` | Miami | | Europe and Middle East | `europe-central-1` | Frankfurt | | Europe and Middle East | `europe-south-1` | Marseille | Premium BGP is available on all ZEC regions. If a region you need is not listed above, Premium BGP is the default path; the console shows only the network types applicable to the region you pick. ### The quick rule * Does an extra 20–40 ms of RTT break an SLO or degrade UX? → **Premium BGP**. * Is the workload dominated by bytes, not by handshake latency? → **Standard BGP**. *** ## How this interacts with billing The network type is picked at creation — it is an attribute of the EIP. Once picked: * **It sets the per-Mbps unit price.** Each network has its own rate card. Premium BGP is at the top; Standard BGP is lower. The billing *method* ([Data Transfer](/welcome/elastic-compute/networking/01-overview/03-data-transfer.md) / [Flat Rate](/welcome/elastic-compute/networking/01-overview/04-flat-rate.md) / [Aggregated 95th](/welcome/elastic-compute/networking/01-overview/05-aggregated-95th.md)) is orthogonal to the network type — each method bills against the rate of whatever network you picked. * **It decides the path and the latency.** Switching meter does not move traffic; switching network type does. * **It is fixed per object.** If you need to change network type, create a new EIP on the other network and migrate. *** ## Rule of thumb * **User-facing, interactive, global?** Premium BGP. * **CDN / VPN / backup / bulk egress?** Standard BGP. Start with the network type that matches the workload's latency tolerance, and let the billing method decide how that bandwidth is metered on top.