FlashEver Storage Assurance Program

Huawei FlashEver leverages core technologies, including unique approximate compression algorithms, to achieve a data reduction ratio of up to 5:1. Huawei FlashEver provides an effective capacity guarantee service to ensure that the amount of application data that can be written to the platform via SAN/NAS protocols is not less than the subscribed capacity. If the SLA commitment cannot be fulfilled, Huawei will provide vouchers to help offset a portion of fees incurred after the subscribed FlashEver service expires.

Huawei FlashEver provides X%~25% additional storage capacity above the contracted amount as buffer capacity at all times. This buffer is essentially a safety net for unexpected traffic demands, ensuring that the service continues to meet its SLA targets.

If buffer capacity is not available and the customer cannot immediately provision needed storage, the provider will waive overage fees or expedite additional capacity at no cost until the buffer is restored.

Huawei FlashEver leverages AI-driven telemetry and analytics to predict and resolve issues proactively, ensuring optimal performance and reliability. The storage platform uses large language models, knowledge bases, machine learning, and more technologies to identify potential risks before impacting the customer (sub-health status or insufficient capacity and performance), identify anomalies within 1 minute, perform intelligent end-to-end analysis for demarcation within 3 minutes, and provide the step-by-step guide to recovery within 5 minutes.

In addition, once an issue happens, the storage platform will automatically notify Huawei technical support team via the Call Home feature, and Huawei will respond within 30 minutes for critical issues, dispatch on-site hardware support within 4 hours, and be available 24/7. If responses or repair times exceed the SLA, the customer will receive service credits and a dedicated review of the incident by the provider's support management.

Huawei FlashEver provides guaranteed streaming throughput for the high-performance tier, with multi-gigabyte throughput per nodes and linear scale-out to over 1,000 nodes. With leading all-flash NVMe arrays and end-to-end performance optimization based on FlashLink features, Huawei FlashEver also provides ultra-low latency for the latency-sensitive workloads, like high-frequency trading and core banking in the financial sector. It ensures sub-millisecond response times, providing consistent, fast data access for applications. It ensures the storage platform can sustain the bandwidth required for large media files, streaming content, or AI/analytics data pipelines without interruption. If the performance falls below the guaranteed level for supported workloads, Huawei will credit the customer and scale the infrastructure (e.g., add performance nodes or increase network bandwidth) at its own expense to meet the required SLA target.

Huawei FlashEver provides an energy-saving guarantee service to help customers reduce energy consumption per TiB of the purchased STaaS services, optimize system OPEX, improve operational efficiency, and meet regional and national carbon emission targets. If the SLA commitment cannot be fulfilled, Huawei will provide vouchers to help offset a portion of fees incurred after the subscribed service expires.

Huawei FlashEver provides ever-new technology refresh for customers during the subscription period, including upgrade to latest hardware, software, and related services. This upgrade can cross different generations, such as an upgrade from OceanStor Dorado V6 to V7, and non-disruptive controller refresh, disk enclosures refresh, etc. And also Huawei FlashEver provides zero hours of planned downtime for upgrades or maintenance (all updates are performed live), to meet the customer's business RPO and RTO SLA target. If any upgrade or maintenance activity causes service disruption, Huawei will issue service credits for the downtime period.

Huawei FlashEver provides efficient and guaranteed virtualization (VM) and data migration (SAN, NAS, and object) services. It ensures a 100% migration success rate, with zero data loss or service interruption for SAN migration, and zero unplanned downtime for NAS and object migration. After a site survey, the migration speed can be guaranteed—typically 3 GB/s for SAN, 800 MB/s for NAS, 650 MB/s for object, and 70 MB/s for virtualization. Besides, Huawei FlashEver provides on-site professional support or tools for partners, backed by Huawei's remote assistance.

Huawei FlashEver enables hybrid multi-cloud deployments by guaranteeing seamless data mobility and replication between on-premises storage and public cloud (or secondary sites). It ensures that data can be moved or synchronized to a cloud environment within set timeframes to enable cloud bursting, disaster recovery, and analytics. The SLA includes a cloud replication tier: a defined Recovery Point Objective (e.g., < 15 minutes) for data replicated to the cloud/secondary site, and a Recovery Time Objective (e.g., < 1 hour) for bringing workloads up in the cloud after an on-prem failure (assuming a failover environment is in place).

Huawei FlashEver helps customers enhance their data resilience and meet compliance goals. It provides a global unified data view, with enriched metadata, user-defined tags, and policies for data classification and sensitivity, preventing non-compliant access and data movement. Additionally, Huawei also provides professional data governance services based on this SLA.

Huawei FlashEver provides unique features like full-mesh architecture, end-to-end DIF, and FlashLink, and guarantees 100% availability (no unplanned downtime in a given period, typically measured annually) to help customers meet high resilience SLA targets.

If availability drops below 100% due to a platform failure during the subscription period, Huawei will work with customers to resolve issues, recover affected data, and provide compensation in the form of service credits or future discounts (e.g., up to 20% of infrastructure cost).

Huawei FlashEver offers real-time ransomware detection and a data recovery guarantee for ransomware protection to minimize customer losses.

Before the event, its HyperDetect feature provides 99.99% ransomware detection rate by automatically analyzing I/O behavior characteristics, notifying the customer, and providing instant responses (like secure snapshots, Air Gap disconnection).

Huawei FlashEver ensures rapid recovery from ransomware or cyber-attacks by providing a clean replacement array shipped by the next business day after an attack, and ransomware recovery expert support within 30 minutes of incident reporting. This includes standby capacity for recovery and secure snapshots that cannot be compromised.

Huawei FlashEver provides LLM-powered intelligent Q&A for customers and partners, supporting deep thinking and long context interactions. It answers over 80% of the typical pre-sales and post-sales storage inquiries like product specification, storage features, solution recommendations, and best practices, with 24/7 availability from anywhere.

Huawei FlashEver helps customers build their own knowledge base for AI applications. This includes data ingestion, cleaning, and conversion into knowledge base for retrieval-augmented generation (RAG), followed by embedding, orchestration, debugging, and release of AI applications. Huawei provides ModelEngine and professional services for customers, with metrics such as data preprocessing speed (e.g. x hours/TB), multi-mode recall rate (e.g. > 90%), and performance (e.g. 10 ms latency, 15,000 QPS) confirmed after a site survey. The SLA can help customers easily manage and build their own knowledge base, better serving AI inference applications.

Huawei FlashEver helps customers improve the AI inference performance and increases token output efficiency by multi-layer KV cache optimization for AI applications. Leveraging the GPU/NPU on-chip memory -> Memory -> AI storage multi-layer KV cache mechanism, it greatly improves the prefill stage efficiency. In typical long-sequence conversational scenarios, over 90% of tokens can be generated by external AI storage, not by computing. Meanwhile, the time to first token (TTFT) can be optimized to less than 3 seconds, and the GPU/NPU utilization can exceed xx% after a site survey. The SLA helps customers optimize the performance and efficiency of their AI inference applications.