5G Markets are developing at a horrendous pace and one thing which is a shared believe across industry is importance of “Edge Networks” to monetize 5G . There are many players in the picture both from Hyperscalers and on-prem vendors and big hybrid cloud players each ready to flood market with “Edge solutions” however fragmentation in this critical component means there will be varying different models of “Edge” consumption which obviously will be a bottleneck to its global adoption .
The purpose of this paper is to share my thoughts around how important is to align on principles and Specs in order to make possible for Telco’s to Mix and Match between different players . First of all when we view Edge from Telecom Operator perspective it means
“Support of Cloud like characteristics near to service consumers including Network boundary (Network Edge) or customer boundary (Customer Edge)”
Below is the summary of those principles that needs to be followed to build a robust “Edge” that can delivers a global scale .
#1:Cloud or Edge
Cloud vs Edge is same like Electricity Generation and Distribution , we know both are important and both need to be modernized over time . With most Apps becoming Cloud Native the core principle is to move those Apps towards Edge which are obviously impractical to host on central cloud due to many factors including regulatory , privacy , latency ,data reduction or technical feasibility aspect . Even in this case the two clouds need to co-exist like Cloud can support Edge clouds to deliver data based solutions , AI use cases or simply use Edge API’s to deliver an E2E use case.
With thousands and may be million of devices at edge the real question is how we can operate it using same cloud type principles we learnt during so many years . It also need to re-use same CI/CD pipelines and test/on-boarding mechanisms they used for Cloud and not just limited to Private but also Public Clouds .
#3:Data and Privacy
Every business is a Digital Business and being secure is now a must have not a nice to have. As a result businesses are assessing their cyber security and resilience posture that are fully compliant with both Data and Privacy concerns , determining where they have gaps and what they need to do about them. A leading Infrastructure provider like Dell Technologies has a valuable role to play, ensuring that security is intelligent, automated and built into everything, everywhere.
Data protection regulations differ between countries and regions (such as the EU). The Edge architecture needs a flexibility to meet Data Protection requirement set by each country and operator
#4:LCM and Day2
What makes Edge a “Ranch” is it is a rugged environment with constraint resources , further tenants are dynamic and commercial model do not enforce a long term commitment something Public clouds always love to have . So ability and build and importantly tear down service on the fly is a must . In addition application mobility strategy that includes QoE, geographical store and privacy policies;
#5:Global Scale and Mobility
One area that is still un-touched is how to open Global Edge owned by 100’s of operators using same API exposure to Developers .In other words Developer can develop Apps in Verizon same way as it can do for Telstra in Australia . It require solve certain issues like
- How to ensure build Edge globally using same principles at least for developers
- How to assure service consistency during Mobility
- Support for roaming where VPLMN can ensure to deliver same service with same QoE something we in Telecom industry has never done
- Able to receive geographical UE mobility events (e.g. when leaving a pre-defined area) from the network or the UE.
Ideally, mobility is handled invisibly to the application’s end-user by the mobile network operator.
#6:Securing the Edge
In explaining Core vs Edge , obviously we expect “Edge” to be universally secure . It require not only “Cloud” all security specs to be followed but also Some specific things related to Edge which are of importance
- Data ingestion and analysis real time e.g to auto lock once a physical intrusion is detected.
- Real time Telemetry based on Streaming and eBPF principles
- Support Silicon root of trust e.g Intel v7
- Residual data clean up (A known issue in industry)
- “Configuration” related CVE poses a biggest threat to Edge security and use of ML/AI to detect and manage “Configuration” and MME (Man made errors) is a must for Edge architectures.
#7:API and Dev to proliferate Edge
Edge require a new way to expose #API and how to define its Specs. Our recent work with in GSMA Edge group may be a good starting point to take #Telco grade view to this #software problem by adressing following .
1. How Different Edge (Public , Private , Hybrid all will expose same end points)
2: How to smoothly tear down an Edge service
3: what are #VNF and #CNF package requirements and how it is different from #ETSI Sol and IFA Specs ? how much we can re-use
4: How to extend #Cloud Infra #CICD to Edge with maximum re-use
5: Monitor Edge resources in real time
6: What are tools and catalog for #Edge , can we re-use #NFVO VNF catalog at the edge and if so how ? as it should be a distributed architecture but managed centrally .
7: How #Dev can ensure #Edge resilience , consider a situation when a 5G and its #Edge resources are down ,
can services can switch consider it was a non H/A site (Again consider $ cost , we can not afford it ).
How #Edge availability can be achieved specially where own Telco sites are not available and need to rely on other Partners and #Telco‘s , how to ensure same service with same QoE for #Dev
Once we imagine the Edge the first and most important feature that comes to mind is resources . Following should be available on Edge Resources
- Inventory for all resources
- Usage of resources via real time monitoring
- Cloud resource reservation mechanism exposure following “need to Know” mechanisms
- Configuring UE App’s to get real time view on customer experience
- Support of resource sharing using USRP and other mechanisms
- Visibility in the Cloudlet locations based on 3GPP LBO mechanisms
- Give fair insights of UE and App Mobility in changing resource scnerios
- Support “Intent” and “Policy” enforcement through Orchestration and Cloud Infra tools whether Edge is a Public or a Private
Ideally the Edge Data Model must give same level of resources exposure and management capabilities to developers of Telco as it can give to Developers outside home country through E-W connectivity and necessary Mobility arrangements
#9:Resource Discovery and Catalog
As developers will build and deploy Applications it becomes increasingly important to have same tools and catalogs that they can use . Normally they can be accessed via E-W API or through MEF and upper layer Orchestration capabilities . The parameters expected are
- Location details
- Infra resources keeping view of stateful workloads
- Acceleration resources like GPU , vPU , NPU and Smart NICS
- QOS profiles both for Provisioning and Monitoring aspects .Monitoring is vital for resources discovery to ensure only valid resources are available for provisioning , this is a complex as it is not just limited to Cloud resources but also need analyze Network topology , transport characteristics etc
- Cost budgets for each AZ , this information is vital (e.g. the use of several small zones, that combined, cover the needed Region and are offered by different partners, instead of a more extensive and expensive zone offered by another partner)
#10:Redefining Edge Application with Telecom Grade
Since there are many Edge and some are infact commercial for long time so it is vital to end the Edge architectures for 5G era with a touch on what is Edge Native for 5G era or simply can we build an Edge application with 5 9’s reliability and a Telco grade that can be benchmarked with same characteristics as provided by large OEM 5G Core CNF’s ? . Such App need to know the App characteristics from user perspective and those App components and architectures needed to manage it .
- Edge application package and artefacts standardization to onboard using uniform capability
- Flavor characteristics consider the fact it can be a different flavors in different Telco’s , Flavor exposure before instantiation is the best Model for a Edge App
- QOS characteristics like Location , Identify , Quality and security
- App state profile to ensure all such components are sinned in a coordinated fashion
- Deployment models needs to be flexible but in certain cases Telco should be ale to select all details of what type of Edge cloud it want to deploy App
- Elasticity profile like how it will scale as per load demand
- Resource migration e.g whether it needs to be auto or steered by customers
- App and Edge cloud coordination is vital , ideally it needs to be de-coupled but we know from experience in Telecom Clouds for many use cases such information coordination between different layers is a key
In a summary we can see that the Edge Cloud and Application with Telecom characteristics is quite different than normal Edge applications available today and mostly offered through public clouds delivery model . It is very important to bring experience from both Telco Clouds and Orchestration/Automation platforms together with NaaS and connectivity framework something Telco’s has been doing for long time .
Experience from Telco Cloud , Orchestration/Automation and NaaS connectivity is the future of Edge Infrastructure and Applications
Final thoughts on how Remote Work will influence Edge Architectures
According to World Economic forum report Sixty-five percent of children starting school today will work in jobs that have not been invented yet .
This re-confirm that there is a different sort of infrastructure required for new era and it will require Edge and EUC (End user compute) and user devices to work in conjunction . Ideally we want to make end device nimble and hence “Edge” solution should off load most heavy lift from end device and still ensuring
- Worker productivity (At least 4X times compared to today)
- Secure work environment with notion that all users will be #WFH primarily accessing the services from remote and hostile access locations that may be an easy Ransom and CVE target
- Data sovereignty and policy , although organizations go remote still IT like same type of control e.g Data sharing , employee accessibility , usability etc
- Auto optimization using ML/AI to ensure infra , tools and software is optimized on a per user level based on his/hear usage trends and not a generalized Infra which is not optimized
- Lastly the security of end user devices through Edge solutions is vital including data analytics ,configuration audits and steer mechanism like auto lock and reboot will be both vital and exciting how Edge solutions need to solve some of the issues of remote Work and its productivity .
I guess this will be an exciting era of Edge and User experience .
Edge + End user productivity solution cohesion is the future of work in post covid-19 world and Infrastructure providers like Dell who operate in both domains is in ideal situation to solve this complex requirement in a simple yet efficient manner