SD-WAN handles whether users notice the outage. The eSLA tier handles how quickly the primary link comes back.
An eSLA tier on NBN Enterprise Ethernet is a contractual commitment to restore a fault within a defined window after it is lodged. The standard tier is 12 hours, 24/7; a 4-hour tier is available as a premium upgrade in most locations. For most Australian multi-site SD-WAN deployments with a properly configured 4G/5G failover link, 12-hour eSLA is the financially rational default — and the right specification is per site, not per estate.
See how SD-WAN changes the decisionWhat eSLA actually covers and what it does not; the 4-hour, 8-hour and 12-hour tier structure; how SD-WAN failover changes the eSLA decision; a four-step per-site decision logic; five scenarios where 4-hour eSLA is justified; SAG, eSLA and Class of Service together; Australian regional eSLA availability; and per-site analysis discipline at procurement. Written for Australian multi-site organisations specifying WAN requirements on NBN Enterprise Ethernet.
The eSLA decision is per site, not per estate.
The 4-hour tier is justified only when the failover link cannot carry the site's critical workload for the additional 8-hour window — or when the site has no failover link at all. The common procurement mistake is specifying 4-hour eSLA across the entire estate as a blanket policy. In a typical 30-site network, four to eight sites genuinely warrant the premium. The remainder do not.
The premium is real money: A$100 to A$300 per site per month, A$2,000 to A$6,000 per month across a 20-site network. That money funds more resilience per dollar when it is spent on better failover devices, higher Class of Service on primary links or per-site analysis discipline than on a shorter primary restoration window for sites that are already handled by failover.
The variables that decide the specification:
- What eSLA covers and what it does not — the eSLA is a restoration commitment, not a prevention commitment. The restoration clock starts at lodgement, not at outage.
- 4-hour vs 12-hour tier comparison — different restoration windows, different cost premiums, different operational impact depending on failover design.
- How SD-WAN failover changes the decision — SD-WAN handles the outage event itself. The eSLA tier then governs how long the site runs on backup, not whether users experience disruption.
- SAG uptime guarantee vs eSLA restoration time — two different parameters in a complete WAN specification. Specifying one without the other is incomplete.
- Class of Service and bandwidth commitment — high CoS guarantees full bandwidth to the network boundary; low CoS is best-effort. Application-aware SD-WAN routing depends on the CoS tier matching application load.
- Per-site cost implications at scale — the premium is material at 20+ sites. Blanket specification across all sites is the most common and most expensive procurement mistake.
eSLA governs restoration time after a fault is lodged. It does not prevent outages.
An eSLA, or Enhanced Service Level Agreement, on NBN Enterprise Ethernet is a contractual commitment by nbn co and the retail service provider to restore the primary fibre service within a defined window after a fault is formally lodged: 24 hours a day, 7 days a week. The standard NBN EE eSLA is 12 hours. A 4-hour eSLA is available as an upgrade at additional cost. In most locations an 8-hour tier also exists.
Understanding what the eSLA clock measures matters for procurement decisions. The eSLA restoration window starts when a fault is confirmed and lodged with nbn co, not when the service goes down. Fault detection, diagnosis and lodgement time adds to the total outage window on top of the eSLA restoration commitment. For a site running without monitoring or a managed service, the actual time from outage to restoration can materially exceed the eSLA figure alone. A well-designed SD-WAN deployment addresses prevention through failover; the eSLA tier addresses restoration speed of the primary link.
What the eSLA does and does not cover:
Once a fault is formally lodged, nbn co and the RSP are committed to restoring the service within the eSLA window. Failure to meet the window typically triggers a rebate on the monthly service fee.
eSLA faults receive prioritised dispatch. The 4-hour tier carries higher escalation priority than 12-hour for nbn co's field teams.
The eSLA does not guarantee the service never fails. It governs restoration speed after failure.
The restoration clock starts after lodgement, not at the moment of outage. Sites without active monitoring may not detect or lodge a fault for minutes or hours after the service drops.
The eSLA applies to the NBN EE primary service. If the site is running on a 4G/5G failover link, the eSLA is still measuring primary restoration. The failover link's performance is separate.
Monthly fee rebates for SLA breaches are typically modest. They do not compensate for revenue loss, operational disruption or reputational damage from an outage.
Three restoration tiers, two that matter for most multi-site decisions.
NBN Enterprise Ethernet eSLA options vary slightly by carrier but typically follow the same three-tier structure. All tiers are 24/7 commitments — not business-hours only.
Standard on NBN Enterprise Ethernet from most carriers. Fault restored within 12 hours of lodgement, 24/7. Included in the base NBN EE price from most carriers without an additional uplift. This is the default tier recommended for most SD-WAN multi-site deployments with 4G/5G failover in place.
Available in most metropolitan areas at a modest uplift over the 12-hour tier. Less commonly specified in procurement documents, which typically compare the two endpoints. Appropriate for sites where the 12-hour window is borderline but the full 4-hour premium is difficult to justify commercially.
Premium tier. Fault restored within 4 hours of lodgement, 24/7. Available on NBN EE in most metropolitan and many regional locations. Typically A$100 to A$300 per site per month more than the 12-hour tier depending on bandwidth tier and carrier. Recommended for sites where the failover link is inadequate or absent, and for mission-critical operational environments.
Typical cost premium for an eSLA upgrade:
- 12-hour (standard) — A$0 to A$50 per site per month
- 8-hour (uplift) — A$50 to A$150 per site per month
- 4-hour (premium) — A$100 to A$300 per site per month
Per-site estimates that vary by carrier, bandwidth tier and location — confirm with the RSP during service qualification. The 8-hour tier is available in most metropolitan areas but is less commonly quoted in multi-site WAN proposals, which typically compare 4-hour and 12-hour directly.
Where the tiers are genuinely different, and where they are not.
The most important framing is that the eSLA tier is a restoration guarantee, not a prevention guarantee. For most multi-site SD-WAN deployments, the SD-WAN failover handles prevention. The eSLA tier handles recovery speed of the primary link.
- Restoration window — up to 12 hours from fault lodgement, 24/7
- Cost premium — standard inclusion; no or minimal uplift over base NBN EE pricing
- Dispatch priority — standard priority in nbn co's field dispatch queue
- Availability — available in all NBN EE footprint locations
- SD-WAN impact — for sites with 4G/5G failover, the 12-hour window is the maximum time on backup; SD-WAN handles the outage itself
- Best fit — most multi-site SD-WAN deployments with adequate 4G/5G failover; standard specification for retail, professional services and distribution
- Restoration window — up to 4 hours from fault lodgement, 24/7
- Cost premium — A$100 to A$300 per site per month additional; material cost at 20+ sites
- Dispatch priority — elevated dispatch priority; faster field crew assignment
- Availability — most metropolitan and many regional locations; confirm during site qualification
- SD-WAN impact — reduces time on the failover link; critical for sites where 4G cannot sustain full load for 12 hours
- Best fit — sites without failover, or where failover cannot sustain critical operations: healthcare imaging, high-transaction EFTPOS, OT environments
The question is not which eSLA tier sounds better. It is: what happens to this specific site in the hours between when the primary NBN link fails and when it is restored?
SD-WAN separates the user experience from the eSLA event.
The most important practical change SD-WAN makes to the eSLA decision is separating the user experience from the eSLA event. In a traditional single-link WAN architecture, an NBN EE fault means the site goes down and stays down until the primary link is restored. The eSLA tier directly determines how long the site is offline.
In an SD-WAN deployment with a 4G/5G failover link, the situation is different. FortiGate SD-WAN monitors link quality continuously and detects degradation or failure within seconds. When the primary NBN EE link fails, traffic automatically switches to the backup path. For most sites, users experience a brief disruption or none at all. The eSLA event is still occurring: the primary link is down, the fault is being lodged, nbn co field teams are being dispatched. But users are working. The eSLA tier now governs only how long the site runs on the backup link before primary is restored.
Investing A$200 per month per site in a 4-hour eSLA upgrade is often a less efficient use of WAN budget than investing A$150 per month in a better-quality 4G/5G failover device that handles the outage itself. Both matter. For most sites, the failover investment provides more resilience value per dollar.
The decision logic with SD-WAN in place:
If no, select 4-hour eSLA. A single-link NBN EE site with no failover is entirely dependent on primary restoration speed. The 4-hour eSLA is the only resilience layer available.
Determine the site's critical bandwidth requirement: ERP sync, EFTPOS transactions, VoIP call capacity, clinical system access. If the 4G/5G link can sustain this for 12 hours, 12-hour eSLA is sufficient. If it cannot, 4-hour eSLA reduces the risk window.
Calculate the operational and financial cost of the site running on the failover link for the 8-hour difference between the 4-hour and 12-hour tiers. If that cost exceeds the annual eSLA upgrade cost of A$1,200 to A$3,600 per site per year, the upgrade pays for itself. If it does not, 12-hour is the rational choice.
In a 30-site network, headquarters and high-transaction sites may justify 4-hour eSLA. Remote depots and low-criticality offices may not. Applying the same eSLA tier across all sites without this analysis is the most common and most expensive mistake in multi-site WAN procurement.
The specific scenarios where the 4-hour tier pays for itself.
The following are not marketing triggers. They are the specific operational conditions where the 8-hour difference between the 4-hour and 12-hour tiers creates measurable risk that outweighs the cost premium. Most sites will not meet these criteria.
A site with only a primary NBN EE connection and no 4G/5G backup is entirely exposed to the primary link's restoration time. A 4-hour eSLA is the only resilience mechanism available.
This should also trigger a conversation about adding a failover link. A well-designed SD-WAN deployment should not rely solely on eSLA tier for resilience.
Healthcare sites with clinical imaging running DICOM over WAN at 20 to 100 Mbps per study. Retail sites with 100+ concurrent EFTPOS transactions. High-volume ERP environments where 4G throughput is insufficient to maintain full operations.
If the site degrades to unusable on 4G within 2 to 3 hours, 4-hour eSLA limits the exposure window.
Some industrial and manufacturing environments have regulatory, latency or security constraints that limit or prohibit 4G as a failover path.
Where the backup option is either absent or severely limited, 4-hour eSLA on the primary link is justified.
Some healthcare, financial services and aged care environments have documented business continuity obligations that specify maximum acceptable downtime periods.
Where a 12-hour primary restoration window would breach a documented BCP commitment, 4-hour eSLA is required regardless of failover availability.
A headquarters site with a single NBN EE feed carrying the full weight of centralised systems, no physical path diversity and a 4G failover link that cannot carry HQ workloads may justify 4-hour eSLA even if a failover link exists.
Apply this analysis per site rather than as a blanket policy across the estate.
SAG and eSLA are two different parameters. Both belong in a WAN specification.
A procurement specification that addresses eSLA tier but not SAG is incomplete. The SAG, or Service Agreement Guarantee, and the eSLA are complementary but measure different things. Specifying both is the correct approach for any multi-site WAN procurement.
- SAG defines how often the service can be down — the Service Agreement Guarantee is an uptime commitment expressed as a percentage. Common options are best-effort with no rebate, 99% uptime, and 99.95% uptime. A 99.95% SAG allows approximately 4.4 hours of cumulative downtime per year. This governs frequency of outage, not restoration speed per event.
- eSLA defines how quickly a fault is restored when one occurs — the eSLA does not reduce the number of faults. It sets the contractual restoration window once a fault occurs. A site can meet its 99.95% SAG and still have a single 4-hour outage in a year — which is within both parameters.
- Class of Service determines bandwidth commitment — High CoS at 100% CIR guarantees the full bandwidth tier to the network boundary. Low CoS is a best-effort service subject to contention. For SD-WAN application-aware routing to work correctly — particularly for Teams, SaaS and ERP traffic — the bandwidth commitment must match the expected application load. An underpowered Low CoS link defeats the purpose of application-aware routing.
Per-site analysis discipline at the specification stage.
A 28-site national distribution organisation was writing a WAN procurement specification for an MPLS-to-SD-WAN migration. The initial draft specified 4-hour eSLA across all 28 sites, based on the assumption that shorter restoration times were always better. Applying the per-site analysis changed the outcome.
- 3 sites — two distribution centres and headquarters — had high-transaction ERP workloads that 4G could not sustain for 12 hours. These retained 4-hour eSLA.
- 4 regional depots with minimal WAN dependency and adequate 4G coverage downgraded to 12-hour.
- The remaining 21 sites were specified at 12-hour with High CoS and 4G failover.
Total underlay cost reduction versus the original all-4-hour specification: approximately A$52,000 per year across the estate, with no meaningful reduction in operational resilience. This is the difference between writing a procurement specification based on assumption and writing one based on per-site analysis.
For most multi-site SD-WAN deployments with 4G/5G failover, 12-hour eSLA is the right default.
For most multi-site Australian organisations deploying SD-WAN with 4G/5G failover, 12-hour eSLA is the right default. Spending the eSLA premium on a better failover link delivers more resilience per dollar for the majority of sites. The reasoning is practical. A well-configured FortiGate SD-WAN deployment detects NBN link degradation within seconds and switches to the 4G/5G backup before most users are aware of a problem. The eSLA tier then governs how long the site runs on backup, not whether users experience an outage. For a site where the 4G link can carry 12 hours of critical workload, the difference between a 4-hour and 12-hour eSLA is invisible to operations.
The A$100 to A$300 per month eSLA premium across a 20-site network is A$2,000 to A$6,000 per month that could fund better failover devices, additional monitoring or higher-CoS primary links — all of which deliver more operational value than a shorter primary restoration window on a link that is already being handled by failover. That is the operating logic managed SD-WAN builds on: the platform and operating model that make 12-hour eSLA workable across most sites.
The 4-hour eSLA is genuinely the right call for specific sites: where failover is absent or inadequate, where the site carries clinical imaging or high-transaction EFTPOS loads that 4G cannot sustain, or where a documented BCP requirement mandates a shorter primary restoration window. The error is specifying 4-hour eSLA across all sites as a blanket policy without running the per-site analysis. In a 30-site estate, applying the analysis correctly typically yields four to eight sites warranting the premium and the remainder on the standard tier — a material cost saving with no operational downgrade.
Optimised resilience per dollar is the benefit. Per-site analysis discipline is the trade-off.
If your current WAN specification applies one eSLA tier across every site, the per-site analysis has not been run — the estate is either paying for restoration windows failover already covers, or leaving critical sites exposed.
Run the per-site analysis →Where Inlight IT works alongside the team:
NBN EE availability, bandwidth tier and Category status per site, 4G/5G coverage testing, and alternative underlay options where fixed-line is constrained.
Decision logic applied site by site against failover capacity, critical workload and BCP requirements, with the cost premium quantified across the estate and the per-site outcomes consolidated into the procurement specification.
High CoS for primary sites running latency-sensitive or guaranteed-bandwidth applications. Medium or Low CoS where best-effort is genuinely adequate for the SD-WAN policy requirements.
4G/5G modem capacity sized against the site's critical workload, not specified as a generic backup.
SLA probes tuned to the underlay, failover thresholds aligned to application requirements, validated under production load before MPLS or sole-link removal.
NBN EE installation lead time of 50 to 80 business days planned in parallel with SD-WAN design, not after design sign-off — with a procurement specification document covering eSLA, SAG, CoS, failover and underlay before tender.
Questions Australian IT and procurement teams ask when specifying WAN requirements.
What is an eSLA on NBN Enterprise Ethernet?
What is the difference between 4-hour and 12-hour eSLA in practice?
Does SD-WAN make the eSLA tier less important?
Should I specify the same eSLA tier across all sites?
What is a SAG and how does it differ from an eSLA?
What is Class of Service on NBN Enterprise Ethernet?
How long does NBN Enterprise Ethernet take to install?
Is 4-hour eSLA available in regional Australia?
Specify the eSLA tier per site before the procurement is issued.
Decide which sites justify the premium before procurement — per-site underlay analysis, eSLA and SAG specification, no platform commitment required.
Book a Network Review