Let's assume I have a set of containers to deploy: what happens if you have a peak of traffic in your PoP? Will my containers be deprioritized ?
GIG Technology value proposition is to deliver private or public cloud services through Edge Cloud. POPs are deployed and operated by GIG Technology Partners or directly by customers. For example: Bancadati (www.bancadati.ch), Kapsch (www.kapsch.net) and Athena Cloud (www.cloudathena.co.uk) are Technology Partners of GIG Technology that deliver public cloud services based on GIG Software. EVS may deploy its own POPs (private or public and offer SaaS services) and/or deploy SaaS services anywhere in the world through GIG Technology Partners.
A GIG Technology customer gets a dedicated Virtual Data Center (VDC) in a POP. Container orchestration is deployed in this dedicated way in a dedicated customer VDC. Concerns related to noisy neighbours are avoided this way. GIG doesn’t over-subscribes the number of subscribers per POP like most Data Center providers currently do. This is a key differentiator of GIG versus traditional Data Centers: no overprovision on bandwidth, on memory or on disk size. GIG provisions CPU by only x 4, or an improvement in performance by a factor of 2 to 3. Customers decide the level of over-provisioning of the POPs directly or or in agreement with the GIG partner.
Connectivity and Latency
Do you guarantee the latency? Which figure? I would say that best use cases of GIG.Tech are low bandwidth (1Gbps) low latency 10ms (FinTech). Correct? I would segment on these 2 dimensions bandwidth & latency. Makes sense?
GIG Technology doesn’t decide the bandwidth. GIG can engineer the latency based on the locations of the nodes and how they are interconnected. This can result in lower overall bandwidth needs. One GIG node can handle 2 X 40 Gbps, so use cases are not limited to low bandwidth use cases. If more bandwidth is needed, more nodes can be added to the POP. So, there is no limitation to the bandwidth. It is a trade-off between locations and capacity per node.
If I want to link my office to Amazon, I have to pay the DirectConnect service (very cheap) and I will pay a BICS subscription to convey my data from my datacenter to Amazon for a guaranteed bandwidth.
How does it work with Gig.Tech ? Do I still need to pay my BICS susbcription to GIG.Tech PoP ? It would indeed work as is expressed above. However the BICS subscription would be reduced, since there is no longer a need for international peering. With the cloud services now running at your premise, in the case of Private Edge Cloud, or at the closest Technology Partner POP, Public Edge Cloud, compute, storage and operational needs are handled locally in an Edge Cloud. Bottomline: expensive peering links are no longer needed with applications deployed near locations where capabilities are needed. Public Cloud runs now locally Edge Cloud POPs.
What are the max specs in terms of connectivity ? I have seen that the nodes were connected through Mellanox & Cisco ? Are these switches 100Gbps ?
As mentioned above, the current configuration is 2 X 40 Gbps (fully redundant, active-active) and even 4 X 40 Gbps is possible. If it is needed for EVS to provide upgrades to > 100 Gbps then this is also possible. We have also attached the specification of the SN2100 Switch to provide all the specifications.
Do you have direct links with CDNs? Which ones?
We are currently in technical and commercial discussions with Atende Software (www.atendesoftware.pl) and can provide more details on this in the near future. We are also open to discuss alternative options with EVS.
Where are the PoPs? Mainly in emerging countries?
GIG Technology offers a Edge Cloud POPs through its Technology Partners in the following countries:
- South Africa
This list is being updated every quarter and partners are being added on a daily basis. GIG Edge Cloud POPs are operated by partners that use the GIG G8 Software to offer public and private cloud services at any location closer to end customers. GIG is also expanding its offering to GIG Play offering SaaS Partners the possibility to re-sell their offerings through Edge Cloud POPs.
What are the max specs of a server (CPU ?, Mem ?) ?
The maximum spec is 1 VM with 256 GB RAM and 56 Vcores
Do you plan to support GPU on the servers ? Do you plan to support FPGA ?
We are evaluating both technologies and will implement in a first phase GPU for Artificial Intelligence (AI) / Machine Learning (ML). Both are roadmap items that can be discussed for acceleration if needed.
Do you support serverless functions (e.g. Lambda on Amazon) ?
We currently don’t support this.
It is mentioned a pay-per-click. What is the model?
GIG only charges based usage. Usage is calculated based on the amount of compute units (2Vcpu 4GB memory), storage units (1 TB each) and transaction units (400 IOPS each) used on a monthly basis. A customer can buy and own an Edge POP directly or work through a GIG Technology Partner. For a detailed simulation of pricing, we recommend working together in order to fine-tune solution (private versus public) and locations of G8 capabilities.
GIG G8 SW requires the below tested and validated form factors.
For storage, do you offer block and/or blob/object storage APIs as S3, Google Storage, etc... ?
We currently offer block storage directly attached as disks to VM’s. We can tune the quality of discs this way and omit the complexity of current centralized cloud offerings where disk types need to defined upfront. GIG allows the same disk to be modulated in line with needs. This lowers costs of discs through a software-based approach (Please also see screenshot of disc modulation features). AWS S3 will be integrated as a standard offering in GIG Play (or GIG’s automation layer) as-a-service. Via GIG Play we can install S3 instances via Minio open-source software (as of January 2019). This is the same for Google storage.
What is the max specs of storage in a node?
The maximum amount of storage in one G8 system is 1.6PB. The HW specifications can be adapted based on commercial needs of a specific use case.
Do you support notifications when files are created, changed, deleted ?
Minio offers auditing for S3 storage in the open-source product. For all other storage devices, GIG offers auditing on resizing, creation and deletion. So, for block storage, we offer full auditing of all changes via audit logs that can be viewed by a customer (Please see below screenshot of audit trail features).
Do you support hot - cold archive ?
GIG uses automatic tiering. All hot data is automatically cached on SSD and all cold on HDD. This doesn’t need to be managed by the customer.
If archive is not supported, are there some automated bridges with archive storage like Glacier?
This is current possible for S3 with Minio. We cannot yet tier at this moment for block devices to lower tiers than HDD. GIG’s block storage does however lower the cost of transferring all the data to Glacier. In addition, the storage system provides anti-bitrot practices, preventing deterioration of data over time.
My datacenter is already hosting containers. What should I install in my datacenter so that I can make use of Gig.Tech as overflow infrastructure??
Nothing needs to be installed in the current DC. By enabling GIG API’s in your pipeline and application orchestration SW, EVS applications can now also run on GIG Edge Cloud infrastructure. GIG provides Ansible modules and Terraform providers for easy integration. Ansible modules are also used to automatically create Kubernetes clusters on a G8.
Do I get something more than Kubernetes to manage the overall infrastructure ? ?
- The overall Edge Cloud infrastructure is managed by GIG Tech or a Partner, so customers need not worry about managing the underlying infra. GIG runs Kubernetes on top of VM’s.
- GIG Play is the automation framework that allows customers to automate the orchestration on top of G8.
- GIG Meneja is a management system that offers a full view of the status of your infrastructure.
Can I set-up some rules somewhere ? Which ones ?
We need to further clarify the question to provide a pointed answer.
What happens when my datacenter is disconnected ? Are there some specific mechanics to automatically scale up the containers on GIG Technology nodes ? I think that could be a good story leveraging the connectivity expertise. Same if connectivity to my datacenter becomes insufficient.
Currently, autoscaling needs internet connection. If the connection is lost, scaling will not happen till the connection is re-established. Loss of connection will not disrupt the entire system, only the auto-scaling functions. In the next design, local scaling without connection will be available.
Are there free tiers like AWS & Google Cloud ?
GIG provides Edge cloud capacity through its Technology Partners. We provide free testing support for VDC capacity to familiarize our customers with our environment and to support them with easy onboarding tools and help. The needs of GIG customers come first. GIG is happy to have an open discussion related to pricing and commercials T&C.
Do you support pricing for reserved instances ?
GIG disrupts current complex pricing practices through a simple pricing based on usage per month as per the Pricing in attachment. Based on our current market intelligence, GIG offers pricing that is 35% more competitive than AWS and Google Cloud for longer term commercial commitments. For private edge cloud (where a company installs a G8 on premise), GIG offers rates that are 25% more competitive than traditional on-premise solutions. A price simulation can be made available to EVS based on compute/storage needs in specific locations.
What are the pricing rules ? Cheap-in / expensive out ?
GIG breaks away from traditional pricing rules of “cheap in” in order to make lots of money on “expensive out”. GIG operates a distributed, scalable Edge Cloud model that disrupts centralized data-hungry cloud models. In the new Edge Cloud paradigm “You bring the network” and “We provide the cloud anywhere”. GIGs business model is not based on margins from HW or network. GIGs business model is based on a distributed cloud paradigm (through partners and directly) where customers are only charged for effective usage as explained under Point 3. GIGs core pricing principles are:
- No hidden cost: License fees are simple and all inclusive
- GIG Play or application automation framework included
- GIG direct customers or Technology Partners only pay for what is used
In which kind of data centers are the nodes deployed ? Tier I II III IV ?
This depends on our partner that runs the G8 POPs. GIG doesn’t host POPs directly. For example Kapch runs a Tier III data center. Athena Cloud runs Tier II data centers. We can provide more details on this based on the locations EVS is exploring.
Is there an impact on prices ?
The hosting provided by GIG distributors is around $0.5/VM/Month, hence the overall impact is low. If EVS is willing to share countries and DC Tier requirements, than we provide a more complete answer. It is also important to stress the availability of GIG nodes in general. G8 nodes are power redundant and network redundant. This enhances protection against DC and network failures.
What are the mechanics to ensure security ??
Several things are in place:
- Dual factor authentication for login
- Secure multi tenant VDC per customer
- Own private VxLan (private network per VDC)
- Own virtual router including firewall, vpn etc…
- Auditing per VDC
- Secure sessions for operations with full text search (google like) for auditing.
Is GIG Technology certified in some way ?
GIG has not yet a certification for security. We can however prioritize this need based on customer requirements.